Pharmaceutical policies: effects of cap and co-payment onrational drug use
Astrid Austvoll-Dahlgren1, Morten Aaserud2, Gunn Elisabeth Vist3, Craig Ramsay4, Andrew D Oxman1, Heidrun Sturm5, Jan PeterKösters6, Åsa Vernby7
1Norwegian Knowledge Centre for the Health Services, Oslo, Norway. 2Statens legemiddelverk, Norwegian Medicines Agency, Oslo,Norway. 3Norwegian Knowledge Centre for Health Services, Oslo, Norway. 4Health Services Research Unit, Division of Applied HealthSciences, University of Aberdeen, Aberdeen, UK. 5Comprehensive Cancer Center, University Medical Center Tübingen, Tübingen,Germany. 6Nordic Cochrane Centre, Rigshospitalet, Dept. 7112, Copenhagen Ø, Denmark. 7Karolinska Institutet, Department ofPublic Health Sciences, Division of International Health (IHCAR), Stockholm, Sweden
Contact address: Astrid Austvoll-Dahlgren, Norwegian Knowledge Centre for the Health Services, Postboks 7004 St. Olavsplass, Oslo,0130, Norway. [email protected].
Editorial group: Cochrane Effective Practice and Organisation of Care Group.Publication status and date: Edited (no change to conclusions), published in Issue 3, 2009.Review content assessed as up-to-date: 2 October 2007.
Citation: Austvoll-Dahlgren A, Aaserud M, Vist GE, Ramsay C, Oxman AD, Sturm H, Kösters JP, Vernby Å. Pharmaceutical policies:effects of cap and co-payment on rational drug use. Cochrane Database of Systematic Reviews 2008, Issue 1. Art. No.: CD007017. DOI:10.1002/14651858.CD007017.
Growing expenditures on prescription drugs represent a major challenge to many health systems. Cap and co-payment (direct cost-share) policies are intended as an incentive to deter unnecessary or marginal utilisation, and to reduce third-party payer expenditures byshifting parts of the financial burden from the insurer to patients, thus increasing their financial responsibility for prescription drugs.Direct patient drug payment policies include caps (maximum number of prescriptions or drugs that are reimbursed), fixed co-payments(patients pay a fixed amount per prescription or drug), coinsurance (patients pay a percent of the price), ceilings (patients pay the fullprice or part of the cost up to a ceiling, after which drugs are free or available at reduced cost), and tier co-payments (differential co-payments usually assigned to generic and brand drugs).
Objectives
To determine the effects of cap and co-payment (cost-sharing) policies on drug use, healthcare utilisation, health outcomes and costs(expenditures).
Search methods
We searched the following databases and web sites: Effective Practice and Organisation of Care Group Register (date of last search: 6September 07), Cochrane Central Register of Controlled Trials (27 August 07), MEDLINE (29 August 07), EMBASE (29 August07), NHS EED (27 August 07), ISI Web of Science (09 January 07), CSA Worldwide Political Science Abstracts (21 October 03),EconLit (23 October 03), SIGLE (12 November 03), INRUD (21 November 03), PAIS International (23 March 04), InternationalPolitical Science Abstracts (09 January 04), PubMed (25 February 04), NTIS (03 March 04), IPA (22 April 04), OECD Publications& Documents (30 August 05), SourceOECD (30 August 05), World Bank Documents & Reports (30 August 05), World Bank e-Library (04 May 05), JOLIS (22 February 06), Global Jolis (22 February 06), WHOLIS(22 February 06), WHO web site browsed(25 August 05).
1Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
We defined policies in this review as laws, rules, or financial or administrative orders made by governments, non-government organisa-tions or private insurers. We included randomised controlled trials, non-randomised controlled trials, interrupted time series analyses,repeated measures studies and controlled before-after studies of cap or co-payment policies for a large jurisdiction or system of care.To be included, a study had to include an objective measure of at least one of the following outcomes: drug use, healthcare utilisation,health outcomes or costs (expenditures).
Data collection and analysis
Two authors independently extracted data and assessed study limitations. We undertook quantitative analysis of time series data forstudies with sufficient data.
Main results
We included 30 evaluations (in 21 studies). Of these, 11 evaluated fixed co-payment, six evaluated coinsurance with a ceiling, fourevaluated caps, three evaluated fixed co-payment with a ceiling, three evaluated tier co-payment, one evaluated ceiling, one evaluatedfixed co-payment and coinsurance with a ceiling, and one evaluated a fixed co-payment with a cap. Most of the included evaluationswere observational studies and the quality of the evidence was found to be generally low to moderate.
Introducing or increasing direct co-payments reduced drug use and saved plan drug expenditures across studies. Patients respondedthrough drug discontinuation or by cost-sharing. Investigators found reductions for life-sustaining drugs or drugs that are importantin treating chronic conditions as well as other drugs. Few studies reported on the effects on health and healthcare utilisation. One studyfound adverse effects on health through increased healthcare utilisation when a cap was introduced in a vulnerable population. Nostatistically significant change in use of healthcare services was found in other studies when a cap was introduced on a drug consideredover-prescribed in a vulnerable population, or following a shift from a two-tier to a three-tier system with increased co-payments fortier-1 drugs in a general population.
Authors’ conclusions
We found a diversity of cap and co-payment policies. Poor reporting of the intensity of interventions and differences in setting,populations and interventions made it difficult to make comparisons across studies. Cap and co-payment polices can reduce drug useand save plan drug expenditures. However, although insufficient data on health outcomes were available, substantial reductions in theuse of life-sustaining drugs or drugs that are important in treating chronic conditions may have adverse effects on health, and as a resultincrease the use of healthcare services and overall expenditures. Direct payments are less likely to cause harm if only non-essential drugsare included or exemptions are built in to ensure that patients receive needed medical care.
P L A I N L A N G U A G E S U M M A R Y
Patient drug payment policies to ensure better use of drugs
Large amounts of health care money are spent on drugs, and these amounts are increasing. Spending more on drugs could mean lessmoney for hospitals, doctors or even other non-health care services. There is also misuse, overuse and underuse of appropriate drugswhich can lead to wasted resources and health hazards. There is therefore a pressure to ensure better use of drugs and to control thecosts of drugs, but without decreasing health benefits.
This review found 21 studies that evaluated policies implemented by governments, non-government agencies and health insurancecompanies to improve drug use or to save (third-party) drug spending or both. Five policies were evaluated in which people pay directlyfor their drugs when they fill their prescription:
caps (prescription drugs are reimbursed up to a maximum amount and then after this amount people have to pay for their drugs),
fixed co-payments (people pay a fixed amount per prescription or drug),
tier co-payments (people pay a fixed amount per prescription or drug which may depend, for example, on whether the drug is a brandname or a generic name),
coinsurance (people pay a percent of the price of the drug),
2Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
ceilings (people pay for part of or full price of the drug up to a maximum amount, for example, for a year, and then people pay no orless money for drugs after that amount).
It is thought that if people have to pay directly for their drugs they may decrease their use of drugs (either overall, or for drugs consideredto be over-prescribed or a less cost-effective alternative than other available treatments); they may buy cheaper drugs; or they may payfor their own drugs, or both.
This review found that cap and co-payment polices can decrease overall drug use and decrease third-party drug spending. But reductionsin drug use were found for both life-sustaining drugs and drugs that are important in treating chronic conditions, as well as in otherdrugs. Although insufficient data on health outcomes were available, large decreases in the use of drugs that are important for peoples’health may have adverse effects. This could lead to an increased use of healthcare services and therefore, overall spending. Policiesin which people pay directly for their drugs are less likely to cause harm if only non-essential drugs are included in these policies orexemptions are built into the policies to ensure that people receive needed medical care.
B A C K G R O U N D
The World Health Organization estimates that “more than half ofall medicines are prescribed, dispensed or sold inappropriately, andthat half of all patients fail to take them correctly. The overuse, un-deruse or misuse of medicines results in wastage of scarce resourcesand widespread health hazards” (http://www.who.int/medicines/areas/rational use/en/index.html).
Simultaneously, growing expenditures on prescription medica-tions represent a major challenge to public policies (Ess 2003;Freemantle 1996; Noyce 2000).
Various pharmaceutical policies are implemented by governmentsand other third-party payers to control expenditures and facilitaterational drug use. This review is part of a series that focuses onthe effects of such policies (Aaserud 2003). In this review, wesummarize what is known from well-designed research on theeffects of policies regarding direct patient payments for drugs.
Most countries or jurisdictions have systems of drug coverage.The systems can be public or private, compulsory or voluntary.Drug coverage is a way of economically securing people’s accessto important drugs, and to spread or diversify the risk of heavyeconomic burdens for people needing expensive drugs. Thus pub-lic health and equity motives are often important for establishingdrug insurance systems.
One downside of drug coverage is the danger of so-called moralhazard (Rice 1994). Drug insurance premiums (either throughtaxes or premiums) are indirect patient payments for drugs. Thesepayments are not related to the specific use of drugs. After the druginsurance premium is paid, the price for drugs is theoretically zerofor patients (unless other bureaucratic costs apply). This may givepatients economic incentives to potentially use more drugs thanthey need. Thus, if a third party pays all the costs, patients can beexpected to have higher utilisation rates (Gross 1994).
By shifting part of the financial burden from the insurer to patients,and thus increasing their financial responsibility for prescriptiondrugs (Gibson 2003; Reeder 1993), direct cost-share policies areintended as an incentive to deter overuse of drugs, use of drugsof limited efficacy or drugs used for conditions where other, morecost-effective treatments are available, and to reduce third-partypayer expenditures. Moreover, direct cost-sharing may reduce theprice of drugs through market mechanisms (Huttin 1994).
Patients are expected to respond to direct payments by:
1. decreasing drug use, either overall or for drugs of limited value;
2. shifting to cheaper drugs; or
3. paying more out of pocket, thus shifting costs from the insurerto patients.
By reducing the financial burden for third-party payers and facil-itating rational drug use, health may be improved by saving re-sources and reallocating them to other health care services (Gibson2005; Reeder 1993). However, the effects of drug policies aremulti-dimensional (Ess 2003). The success of a drug policy shouldnot only be measured by its effect on drug use and drug costs, butalso by its effects on health, the use of healthcare services, and costsfor patients as well as the insurer. A too-restrictive drug policy mayhave unintended consequences. For example, a shift of cost frominsurer to consumer in low-income or other vulnerable popula-tions may lead to discontinuation of necessary drugs, which maycause deterioration of health, increased healthcare utilisation andexpenditures for patients as well as insurers.
Categories of direct patient drug paymentpolicies
3Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Direct patient drug payment policies are diverse, relying on dif-ferent mechanisms. They vary with respect to the drugs that areincluded, the patient groups that are targeted, their intensity (sizeof co-payments), exemptions, enforcement and the units on whichthe payments are imposed (prescriptions, items, doses, expendi-tures). A number of different terms and definitions are used todescribe these various policies. We have pragmatically categorisedthem into five main groups, based on the policies’ intentions andmechanisms in a neutral regulatory environment. In real life, drugpolicies are implemented separately, or combined with other drugpolicies in complex settings involving other health polices and reg-ulations.Caps
A cap is a prescription limit that allows unconstrained (zero-price)purchase by the patient of a certain number or volume of pre-scriptions, drugs or doses in a defined period of time, and imposesfull patient payment of the market price after the limit is reached.Caps may be applied to less cost-effective prescription drugs withthe aim of reducing their use. Alternatively, they may be appliedto all prescription drugs covered by the plan, with the aim of in-fluencing patients and physicians to prioritise the use of the mostimportant drugs. Another intention is to reduce overall drug use,and to reduce plan expenditures by shifting some of the financialresponsibility from the insurer to the patient. Multi-drug users areparticular vulnerable to this policy, since they are most likely toneed prescriptions beyond the cap, especially when life-sustainingor other important drugs are included. Since many multi-drugusers are elderly or have chronic conditions, this policy may haveunintended effects on vulnerable populations. Other patients withdrug use not exceeding the cap limit will have full drug coverage.Fixed co-payments
A fixed co-payment is a payment by the patient of a fixed amountper drug or prescription. This is also called a prescription charge,consumer charge, prescription fee, patient fee or cost-sharing. Theaim of a fixed co-payment is to reduce overall drug expendituresand utilisation. Since co-payments are the same for every prescrip-tion or for the whole drug group, patients’ co-payments are iden-tical for both brand and generic drugs. Therefore fixed co-pay-ments do not provide incentives to choose cheaper substitutions,in contrast to coinsurance or tier co-payments.Coinsurance
Coinsurance is a co-payment based on a set percentage of the priceof the prescription or drug, rather than a fixed fee. This is some-times also called co-payment or cost-sharing. Coinsurance pro-vides an incentive for patients to choose cheaper drugs. Otherwise,co-insurance is similar to fixed co-payment, in that the policy aimsto reduce overall drug expenditures and utilisation.Ceilings
A ceiling is a maximum contribution policy where patients mustpay either full cost or part of the cost (when combined with a fixedco-payment or coinusurance) up to a certain amount per definedperiod of time. After the ceiling is reached by the patient, drugs are
free or available at reduced cost for the patient. This is sometimescalled a maximum contribution, a deductible, or a safety net. Amaximum contribution policy is a more friendly policy towardspatients with chronic conditions and multi-drug users, who aremost likely to exceed the ceiling. Other patients with only a fewprescriptions during the period will not benefit. The policy maybe hard on low-income populations, depending on how high theceiling is set. Ceiling policies may also potentially introduce ahoarding effect after the ceiling is reached.Tier co-payments
Tier co-payment structures commonly assign generic drugs thelowest co-pay (first tier), and higher co-payments for brand drugs(second tier). If the structure is a three-tier system, the second tieris then assigned to preferred brand drugs and the third tier to non-preferred brand drugs. Tier co-payments are also called incentive-based formularies or differential co-payments, and encourage con-sumers through financial incentives to choose products that areassumed to be more cost-effective for the insurer. With this policy,one would expect reductions in the total drug use, compared tono co-payment, and a shift from expensive brand drugs towardscheaper second- or first-tier drugs. This shift is expected to furtherreduce expenditures for the insurer, but may also increase expen-ditures for patients unwilling to substitute for cheaper drugs or ifsubstitution is not possible. Tier structures may also give insurersmore bargaining power to negotiate rebates with drug manufac-turers. In this way, tier structures have similar features to formu-laries and reference drug pricing systems (Aaserud 2006).Modifying factors
The impact of implementing direct patient payments for drugsdepends on the intensity of the intervention, the type of payment(fixed co-payment, coinsurance, etc.), the length of the period therestriction encompasses (for caps and co-payment ceilings) andthe units the payments are imposed on (prescriptions, items, dosesor expenditures). How patients react to these policies also dependson price elasticity.The price elasticity for a drug is the percentage change in its con-sumption related to one percentage change in the price or chargethat patients pay for that drug (Domino 2003; Huttin 1994;Johnston 1991). This is a measure of how sensitive drug con-sumption is to changes in drug prices, and indirectly to changesin co-payments. High-income patients or patients in greater needof drugs are expected to be less sensitive to co-payments (Smith1992). The effect on their drug use may be unchanged, if theyvalue the drug as more important than the burden of cost sharing,and choose to pay the increased drug expenditures to sustain theirdrug use. However, several other factors may modify the impactof direct payment policies (See Additional Table 1):- drug groups included in the policy;- vulnerable populations;- enforcement;- information provided to patients and providers;- exemptions.
4Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Direct payment by patients for drugs is controversial, because in-creased cost sharing for drugs may present a financial barrier topoor households or to patients with chronic conditions who needa high volume of pharmaceuticals (Reutzel 1993; Soumerai 1990;Thomson 2004). Low-income populations may be particularlyvulnerable to co-payments because they are also more likely to besick (Adams 2001). Other vulnerable groups can include pregnantwomen, children and the elderly (Rice 2004; Soumerai 1987).Exemptions are often made to protect the disadvantaged (Ess2003; Haaijer-Ruskamp 2002; Mossialos 2004). However, exemp-tions also reduce the potential savings for the insurer. Exemp-tions can be embedded into the policy for certain drugs or patientgroups based on specific criteria, such as disease, age or income(Gross 1994). One mechanism for providing exemptions is priorauthorisation, where reimbursement for a restricted prescriptionmay be permitted for certain patients fulfilling set criteria (Martin1996). Another loophole for co-payment per prescription policiesis allowing prescribers to increase the volume (doses) allowable perprescription or prescription item (Soumerai 1994).The enforcement of the policy may also have an impact on the ef-fect of the policy. Physicians or pharmacists may decide to exemptpatients from the policy; however, they may then be accountablefor the co-payment instead of the patient.How involved patients are in decision making, and the informa-tion provided to prescribers or pharmacists are important factors.If physicians rather than patients determine which drugs are pre-scribed, they might not be as sensitive to higher patient payments.The extent to which they are informed about the price of drugs,drug substitution possibilities and the patient’s ability to pay canalso affect the impact of direct payment policies. If prescribers arenot well informed, the use of drugs may not change and the eco-nomic burden on patients may increase.The information provided to patients is also important, becausetheir decisions may depend on their knowledge about their healthand the effects of different drugs. Critics argue that because mostpatients lack relevant knowledge, they may be at risk of relinquish-ing drugs with important health effects instead of drugs that areless essential (Levy 1992; Lexchin 2002).
Other cost-share policies not included in thisreview
The design, mechanisms and intended effects of different cate-gories of pharmaceutical policies overlap. For example, policiesthat set reimbursement prices, like reference pricing (Aaserud2006), are similar to co-payment policies, in that both influencewhat the third-party payer and patients pay for the drugs. Thedifference is that patients can choose to use the reference drug andthus not have to pay a reference premium, whereas with co-pay-ments patients have to pay a portion of the cost regardless whichdrug they use within a drug group.
Another related group of polices is formularies. These policies de-fine a list of drugs that are reimbursed or recommended by theinsurer. If the physician prescribes non-formulary drugs or the pa-tient chooses to purchase drugs outside this list, the cost must thenbe fully or partly covered by the patient. Formularies range fromopen informational lists, to partially closed lists restricting selecteddrugs, to closed formularies. Instead of including drugs on thelist based on criteria of efficacy, safety and cost-benefit measures(Dewa 2003; Jang 1988), tier structures use differentiated pay-ments to facilitate substitution from brand to generic drugs. Al-though similar to tier co-payments in many ways, formularies donot offer patients the same financial responsibility. Nevertheless,the policies overlap and are often combined. For example, tier co-payments are often referred to as incentive-based formularies ormulti-tier formularies. Likewise, formulary polices are often com-bined with different types of co-payments.Reference pricing and formulary policies are addressed in separatereviews (Aaserud 2003).
Other reviews
We have identified no systematic reviews of cap and co-paymentpolicies. Although a number of literature reviews have been pub-lished, these were limited in scope or only addressed one group ofdirect payment policies (Adams 2001; Freemantle 1996; Gibson2003; Ginsburg 1973; Gleason 2004; Griffin 1996; Gross 1994;Haaijer-Ruskamp 2002; Harten 2004; Huttin 1994; Levy 1992;Lexchin 2002; Lyles 1999; Nair 2004; Narine 1997; Reeder 1993;Rice 1994; Rice 2004).Our aim in this review is to support informed decisions aboutpharmaceutical policies and to guide future evaluations by prepar-ing an up-to-date, comprehensive summary of what is known fromwell-designed research about the effects of alternative cap and co-payment policies for improving rational (appropriate and efficient)drug use.
O B J E C T I V E S
To determine the effects of cap and co-payment policies on rationaldrug use.
M E T H O D S
Criteria for considering studies for this review
Types of studies
5Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Healthcare consumers and providers within a large jurisdiction orsystem of care. Jurisdictions could be regional, national or interna-tional. Studies within organisations, such as health maintenanceorganisations, were included if the organisation was multi-sitedand served a large population.
Types of interventions
Policies that regulate out-of-pocket payments for drugs by patients,including increases and decreases in the amount paid directly bypatients, limits on the amount paid by patients, and limits on theamount reimbursed, including caps, fixed co-payments, coinsur-ance, maximum co-payment ceilings and tier co-payments.We defined policies in this review as laws, rules or financial oradministrative orders made by governments, non-government or-ganisations or private insurers. We excluded interventions at thelevel of a single facility. However, if an intervention was a pilotstudy of a proposed policy, but evaluated in a single facility servinga large population, we would have included it.
Types of outcome measures
To be included, a study had to include an objective measure of atleast one of the following outcomes.1. Drug use2. Healthcare utilisation3. Health outcomes4. Costs (expenditures), including drug expenditures and prices,other healthcare and policy administration expendituresDrug groups reported on in studies included in this review are of-ten categorised as “essential” (usually drugs that are life-sustainingor drugs that are important in treating chronic conditions) or “dis-cretionary/less essential” (often referring to drugs that are symp-tom relieving, but also to drugs that are considered to be over-pre-scribed or a less cost-effective alternative than other available treat-ments). Which drugs are included in these respective categoriesvaries over time and across studies. To provide a better overviewof which drugs were included in each study, we have summarisedthe categorisation of drugs in Additional Table 2.
Search methods for identification of studies
The search to identify studies for this review was initially done asa part of a series of reviews on the effects of all pharmaceuticalpolicies (Aaserud 2006).
Initial broad search for studies of pharmaceutical
policies
We developed the search strategy without language restrictions.The following databases were searched:Effective Practice and Organisation of Care Group Register, Ide-alist database searched 22/08/03MEDLINE Ovid, 1966 to JuneWeek 1 2003, searched 18 June03EMBASE Ovid, 1980 to 2003 Week 23, searched 18 June 03CENTRAL, The Cochrane Central Register of Controlled Trials,Ovid, searched 15 October 03CSA Worldwide Political Science Abstracts from 1975-present,searched 21 October 03EconLit WebSPIRS from 1969-present, searched 23 October 03SIGLE, System for Information on Grey Literature in Europe,WebSPIRS from 1980-2003/06, searched 12 November 03INRUD, International Network for Rational Use of Drugs,searched 21 November 03PAIS International, Public Affairs Information Service, Web-SPIRS from 1972-2003/07, searched 23 March 04International Political Science Abstracts, WebSPIRS from 1989-2003/12, searched 09 January 04NHS EED, National Health Services Economic EvaluationDatabase, CRD, searched 20 February 04PubMed searched 25 February 04 for relevant journals not indexedin MEDLINENTIS, National Technical Information service from 1964-present,searched 03 March 04IPA, International Pharmaceutical Abstract, WebSPIRS from1970-2003/12, searched 22 April 04The Health Management Information Consortium (HMIC)database was tested and found not to be useful for this review.In addition the following web sites and databases were searched:OECD (Organisation for Economic Co-operation and Develop-ment) Publications & Documents, searched 30 August 05SourceOECD, searched 30 August 05World Bank Documents & Reports, searched 30 August 05World Bank e-Library, searched 04 May 05JOLIS, The Library Network serving theWorld Bank Group andIMF, searched (22 February 06)Global Jolis, online catalogue for theWorld Bank Country OfficePIC/Libraries, searched (22 February 06)WHO (World Health Organisation), browsed (22 February 06)WHOLIS, the WHO library database, searched (22 February 06)The MEDLINE search strategy was mainly developed using re-views cited in the background section of the protocol and theirreferences. The strategy includes terms for the followingcategories of interventions:Regulation and classification (licensing) policiesPatent and profit policiesMarketing policiesPolicies that regulate the provision of drug insurance
6Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Policies that determine which drugs are reimbursedRestrictions on reimbursed drugsPrescribing policiesPricing and purchasing policiesRegulation of salesCap and co-paymentsPatient informationWe used a modified version of the EPOC search strategy method-ology filter to limit the MEDLINE strategy to randomised trials,controlled trials, time series analyses and controlled before-afterstudies.Search strategies for most of the other databases were developed onthe basis of the MEDLINE strategy. We screened the reference listsof all of the relevant reports that we retrieved. We contacted authorsof relevant studies, relevant organizations and discussion lists toidentify additional studies, including unpublished and ongoingstudies.
Further specific searches for studies of cap and co-
payment policies
We performed a subsequent search for studies of cap and co-pay-ment policies before publishing this review. Search strategies weredeveloped for MEDLINE and EMBASE, based on relevant partsand yields of the initial broad search strategy.MEDLINE OVID (1966 to February Week 3 2006), searched(01 March 2006), updated again (29 August 07)EMBASE Ovid (1980 to 2006 Week 08), searched (01 March2006), updated again (29 August 07)ISI Web of Science, searched (09 January 07) for cited key refer-encesEffective Practice and Organisation of Care Group Register (06September 07)Cochrane Central Register of Controlled Trials (27 August 07)NHS EED (27 August 07)The MEDLINE OVID search strategy used both MeSH termsand text words (tw):1. *Cost Sharing/2. *“Deductibles and Coinsurance”/3. *Health Benefit Plans, Employee/4. *Capitation Fee/5. *Fees, Pharmaceutical/6. *“Fees and Charges”/7. *Prescription Fees/8. *Single-Payer System/9. (cost? adj2 (share or shared or sharing)).tw.10. (deductible? or coinsurance or co insurance).tw.11. benefit plan?.tw.12. capitation?.tw.13. (cash adj1 pay$).tw.14. ((charg$ or fee? or direct pay$ or direct contribut$) adj3 (pa-tient? or prescrib$ or prescrip$ or pharmaceutic$ or pharmacy or
pharmacies or dispens$)).tw.15. (pocket adj3 pay$).tw.16. (copay$ or co pay$).tw.17. ((limit$ or cap$ or restrict$ or reduc$ or regulat$) adj3 (pre-scrib$ or prescrip$ or reimburs$)).tw.18. (tier or tiered system? or multitier$ or onetier$ or twotier$ orthreetier$).tw.19. single pay$.tw.20. or/1-1921. exp *Pharmaceutical Preparation/22. *Prescriptions, Drug/23. *Drug Utilization/24. *Drug Costs/25. (drug or drugs or pharmaceutic$ or medicines or medicament?or medicat$ or prescrib$ or prescrip$).tw.26. or/21-2527. State Health Plans/28. State Medicine/29. Government Programs/30. National Health Programs/31. Medicaid/32. Medicare/33. Health Maintenance Organizations/34. Health Policy/35. Health Care Reform/36. Policy Making/37. Legislation, Drug/38. (regulat$ or requirement? or restrict$ or monitor$ or control$or legislation? or law? or act? or policy or policies or reform$ orsystem? or plan$ or program$ or strateg$ or state$ or government?or medicaid or medicare or health maintenance organi#ation? orhmo?).tw.39. or/27-3840. randomized controlled trial.pt.41. controlled clinical trial.pt.42. intervention studies/43. experiment$.tw.44. (time adj series).tw.45. (pre test or pretest or (posttest or post test)).tw.46. random allocation/47. impact.tw.48. intervention?.tw.49. chang$.tw.50. effect?.tw.51. evaluation studies/52. evaluat$.tw.53. comparative studies/54. (randomized or randomised).tw.55. (random$ adj1 (allocat$ or assign$)).tw.56. or/40-5557. Animals/58. Humans/
7Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
59. 57 not 5860. letter.pt.61. comment.pt.62. editorial.pt.63. or/60-6264. 20 and 26 and 39 and 5665. 64 not (59 or 63)The search strategies for the other databases and websites are pro-vided in the appendices (Appendix 1; Appendix 2; Appendix 3;Appendix 4; Appendix 5; Appendix 6; Appendix 7; Appendix 8;Appendix 9; Appendix 10; Appendix 11; Appendix 12; Appendix13; Appendix 14; Appendix 15; Appendix 16; Appendix 17; Ap-pendix 18; Appendix 19; Appendix 20).
Data collection and analysis
Two authors (of AA, MOA, GEV, JPK and HS) independentlyreviewed all of the search results, abstracts and reference lists ofrelevant reports. The full text of potentially relevant reports wasretrieved (if one or both authors thought it was potentially rele-vant) and two (of the above) authors independently assessed therelevance of those studies and the limitations of included stud-ies. The lead author (AA) extracted data from included studies incollaboration with one other author (MOA, GEV, CR ADO orÅV). For all the steps in the above process, the authors resolveddisagreements by discussion, when necessary including anotherauthor (ADO or CR).
Included study limitations
We used the standard criteria recommended by EPOC to assessthe methodological limitations of studies (risk of bias) included inEPOC reviews (EPOC 2002).The criteria for RCTs and CCTs were as follows.1. Concealment of allocation.2. Baseline measurement of outcomes.3. Follow up of professionals.4. Follow up of patients.5. Intention-to-treat analysis.6. Blinded assessment of primary outcomes.7. Reliable primary outcomes measures.8. Other risk of bias.The criteria for CBA studies were as follows.1. Baseline measurement of outcomes.2. Baseline characteristics of studies using second site as control.3. Follow up of professionals.4. Follow up of patients.5. Reliable primary outcomes measures.6. Blinded assessment of primary outcomes.7. Protection against contamination.8. Other risk of bias.
We used the EPOC definition of RCT, CCT, CBA and ITS stud-ies. For ITS studies the definition is: “The study must have a clearlydefined time of intervention AND must have at least three datapoints before and three data points after the intervention.” We alsoconsidered designs where there was a control ITS group. ControlITS designs are conceptually similar to CBA design, but the addi-tion of multiple time points pre- and post-intervention decreasesthe likelihood of secular change bias. ITS studies usually have onlyone data item at each point in time (e.g. number of hospitalisedcases). However, in this review there were studies that had ITSdata for many individual patients (i.e. each individual contributeddata to every point in time). Such designs usually produce a higherlevel of evidence than a simple ITS, because between-patient vari-ability can be modelled as well as within-patient variability, result-ing in a study with substantially more power than a simple ITS.The criteria for protection against bias are the same as for a simpleITS, except that the appropriate methods of analysis differ (e.g.repeated measures anova, generalised estimating equations or ran-dom-effects models). We have called these studies repeated mea-sures (RM) studies. As with an ITS study with a control group,RM studies can also have a control group consisting of controlpatients repeatedly observed over time.Based on experience with two previous systematic reviews (Davey2005; Grilli 2002), the statistical editor of EPOC, who is also a co-author of this review (CR), suggested minor revisions to the EPOCcriteria for ITS (and RM studies). These consisted of definingreanalysed studies as meeting the ’analysed appropriately’ criteria,and allowing studies that had at least 12 monthly data points preand post to meet the ’reason for number of data points’ criteria (the12 monthly points allowed the possibility of investigating seasonaleffects). These criteria more accurately reflected the chance of biasin the study effect sizes. We therefore used the following criteria:1. The intervention was independent of other changes (protectionagainst secular changes). This was ’MET’ if there were compellingarguments that the intervention occurred independently of otherchanges over time and the outcome was not influenced by otherconfounding variables/historic events during study period.2. Data were analysed appropriately. This was ’MET’ if autore-gressive integrated moving average (ARIMA) models were usedOR time series regression models were used to analyse the dataand serial correlation was adjusted/tested for OR reanalysis per-formed, OR any other time series method that was judged to haveaccounted for autocorrelation or secular trends (there are manydifferent appropriate techniques possible; the above are just themost common).3. Reasons for number of data points were given. This was ’MET’ ifdata for 12 months (or more) pre- and post-intervention were usedOR sample size calculation performed OR reason for the numberand spacing of data points is given (should be stated explicitly whythe number and spacing of data points were chosen if the othertwo criteria are not met).4. Shape of the intervention effect was pre-specified. This was
8Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
’MET’ if point of analysis was the point of intervention OR arational explanation for the shape of intervention effect was givenby the author(s). Where appropriate, this should include an expla-nation if the point of analysis was NOT the point of intervention.5. Intervention unlikely to affect data collection (protectionagainst detection bias). This was ’MET’ if it was reported that in-tervention itself was unlikely to affect data collection (for example,sources and methods of data collection were the same before andafter the intervention).6. Blinded assessment of primary outcome(s). This was evaluatedas protection against detection bias. This was ’MET’ if the authorsstated explicitly that the primary outcome variables were assessedblindly OR the outcome variables were objective, e.g. length ofhospital stay, drug levels as assessed by a standardised test.7. Completeness of data set. This was ’MET’ if the data set covered80-100% of total number of participants or episodes of care in thestudy.8. Reliable primary outcome measure(s). This was ’MET’ if two ormore raters with at least 90% agreement or kappa greater than orequal to 0.8 OR the outcome was obtained from some automatedsystem, e.g. length of hospital stay, drug levels as assessed by astandardised test.9. Other risk of bias.For controlled ITS (CITS) and controlled RM (CRM) studies,the time series part of the studies was assessed independently fromthe control part, using the above described criteria for ITS andRM studies. The control series part of the study was assessed usingthe CBA criteria above. If the control part had a high risk of bias,it was not included and the study was classified as ITS or RM,otherwise the control data were used as a control in the review.Overall limitations for each main outcome within each study wereassessed by each of the data extractors using the following guide-lines:1. no serious limitations = low risk of bias = all criteria scored as’met’;2. some limitations = moderate risk of bias = one or two criteriascored as ’not clear’ or ’not met’;3. serious limitations = high risk of bias = more than two criteriascored as ’not clear’ or ’not met’;4. fatally flawed = study results that we believed to be untrustwor-thy, based on an overall judgement of the risk of bias in the study,based on all of the criteria used to assess the risk of bias.We excluded studies rated as ’fatally flawed’ from the review, butlisted them, with the reason for exclusion, in the table ’Character-istics of excluded studies’. We used some setting-dependent judg-ment (i.e. judgment dependent on knowledge of the setting inwhich a study was done) when assessing overall limitations. Wheresetting dependent judgment has been used, we have provided ex-planations in Additional Table 3 and Table 4.
Data Extraction
We extracted the following additional information from includedstudies, using a standardised data-extraction form.1. Type of study (randomised trial, non-randomised trial, repeatedmeasures study, interrupted time series, controlled before-after).2. Study setting (country, key features of the healthcare system andconcurrent pharmaceutical policies).3. The sponsors of the study.4. Characteristics of the participants (consumers, physicians, prac-tices, hospitals, etc.).5. Characteristics of the policies.6. Main outcome measures and study duration.7. The results for the main outcome measures.We included all outcomes that met the inclusion criteria. However,if the study presented results on the same outcome several times(perhaps just using different units), or across a large number ofdrug groups, we chose what we considered the most importantoutcomes in each of the four outcome groups (drug use, health,healthcare utilisation and expenditures), either as specified by theauthor or based on discussions among the authors. We includedadditional outcomes based on the judgement of two authors if theyprovided a better insight into how a policy worked or didn’t workas expected, or if they provided an explanation of the mechanismsor modifiers of the intervention. We did not base decisions aboutwhich outcomes to include on the direction or size of effect, orwhether a finding was statistically significant.We prepared tables for each subcategory of intervention, includ-ing the following information: study identification; characteristicsof the intervention; drug use; healthcare utilisation; health out-comes and expenditures. We also described potential mechanismsthrough which the policies were intended to affect drug use andexpenditures, and postulated mechanisms for other effects, bothintended and unintended. We also briefly listed and described im-portant policy options for which no evaluations were found. Ourconfidence in the available estimates of effects was graded usinga modification of the approach recommended by the GRADEWorking Group (GRADE 2004). When grading the quality ofevidence, we initially graded ITS and RM studies as ’Moderate’quality, and CBA studies as ’Low’ quality. This reflects our judge-ment that interrupted time series and repeated measures studiesprovided more compelling evidence than CBA studies in evalu-ating policies. When a majority of ITS/RM studies evaluated anoutcome or there was an equal number of both ITS/RM and CBAevaluations, the evidence was not graded down. When the ma-jority of CBA studies reported on an outcome, the evidence wasgraded down. The GRADE quality scores are High, Moderate,Low and Very low. If only one study was included in a policy group,it was not presented in a summary of findings table. Likewise, ifthe mechanisms of an intervention were considered to be substan-tially different (for example, reduced co-payment versus increasedco-payment) from the other policies in a particular group, or ifno data were available, we only reported the study in the ’Results’section, and did not include it in a summary of findings table.
9Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
We considered the following potential explanatory factors for het-erogeneity: differences in the characteristics of the policies; differ-ences in the settings; and differences in study limitations. How-ever, there were generally not sufficient comparisons for similaroutcomes across studies to allow for meaningful exploration ofheterogeneity.In addition, we attempted to identify important factors that mightbe taken into consideration by anyone contemplating implement-ing any of the policy alternatives, including: possible trade-offs (ofthe expected benefits versus harms and costs); short- versus long-term effects; indications and contraindications for when the poli-cies might be used; limitations of the available evidence; and otherimportant factors that might affect the translation of the availableevidence into practice in specific settings.ITS and RM studies
The preferred analysis method for ITS and RM studies was ei-ther a regression analysis with time trends before and after theintervention, which adjusted for autocorrelation and any periodicchanges; or ARIMA analysis or other techniques that adjusted forautocorrelation and secular trends. The results for the outcomesshould be presented as changes along two dimensions: change inlevel and change in slope. Change in level is the immediate effectof the policy and is measured as the difference between the fittedvalue for the first post-intervention data point (one month afterthe intervention), minus the predicted outcome one month afterthe intervention, based on the pre-intervention slope only.Change in slope is the change in the trend from pre- to post-inter-vention. It reflects the ’long’-term effect of the intervention. Sincethe interpretation of change in slope could be difficult, we choseto present the long-term effects similar to the way we calculatedand presented the immediate effects. We presented the effects afterhalf a year as the difference between the fitted value for the sixth-month post-intervention data point (half a year after the interven-tion), minus the predicted outcome six months after the interven-tion, based on the pre-intervention slope only. We reported theeffects after one and two years in the same way when they weremeasured. For drug expenditures, we also calculated the savingsafter a half, one and two years as the area between the predictedand actual expenditures curves.Given that policy changes are often announced some months priorto official implementation, we defined a transition phase as sixmonths from the official announcement of the policy. If the in-cluded ITS and RM studies stated a different transition phase,we used the studies’ definition. All results excluded the transitionphase data.If papers with an ITS design did not provide an appropriate analy-sis or report of the results (Cromwell 1999; Martin 1996), but pre-sented the data points in a scannable graph or in a table, we (CR)reanalysed the data using methods described in Ramsay 2003. Thefollowing segmented time series regression model was specified:Y(t) = B0 + B1*Pre-slope + B2*Post-slope + B3*intervention + e(t)where Y(t) is the outcome in month t. Pre-slope is a continuous
variable, indicating time from the start of the study up to the lastpoint in the pre-intervention phase and coded constant thereafter.The post-slope was coded as 0 up to and including the first pointpost-intervention, and coded sequentially from 1 thereafter. Theintervention was coded as 0 for pre-intervention time points, and1 for post-intervention time points. In this model, B1 estimatesthe slope of the pre-intervention data, B2 estimates the slope ofthe post-intervention data and B3 estimates the change in level ofoutcome as the difference between the estimated first point post-intervention and the extrapolated first point post-intervention, ifthe pre-intervention line was continued into the post-interventionphase. The difference in slope is calculated by B2-B1. The errorterm e(t) was assumed to be first order autoregressive. We calcu-lated confidence intervals (95%) for all effect measures.In a repeated measures design, the data are repeated outcome mea-sures from many individual patients. For studies with this design,we used the results reported in the original papers, since any re-analysis would underestimate or overestimate the standard error ofthe effect sizes because it was not possible to adjust for correlationoccurring within individuals..However, for studies with adequate analysis, but where no appro-priate effect sizes were reported (Blais 2002; Tamblyn 2001), bothrepeated measures and ITS designs that graphically displayed theconfidence intervals for the projected post-intervention estimates(based upon the pre-intervention data only) were reanalysed. Thewidth of the newly derived effect size confidence interval was as-sumed to be the same as the width of the projected confidenceinterval in the original study analysis. Although imperfect, thisapproach should only marginally under- or overestimate the stan-dard error of the effect sizes. Given the pragmatic nature of thisapproach, we have highlighted the reanalysed results where weperformed this.For studies that only reported the absolute change (Donnelly 2000;McManus 1996; Motheral 2001; Nelson 1985 (see Reeder 1985)),we estimated the relative change in level from the expected level oneobservation post-intervention. We calculated the expected levelas the intercept plus the pre-policy trend times the number ofobservations up to the first time point post-intervention.Whilst we chose to use a relatively simple method of reanalysisof interrupted time series, we accept that the precision of the es-timates from the reanalyses might be biased (upwards or down-wards). It is for this reason that we emphasise the effect size andnot the statistical significance of the effect size of ITS studies inthe ’Summary of results’ tables.CBA studies
For CBA studies, we reported adjusted relative effects. For di-chotomous outcomes we have reported, if possible, the relativerisk, adjusted for baseline differences in the outcome measure; i.e.the relative risk post-intervention/relative risk pre-intervention.For continuous variables we have reported, if possible, the rela-tive change, adjusted for baseline differences in the outcome mea-sure; i.e. (the absolute post-intervention difference between the
10Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
intervention and control groups - the absolute pre-interventiondifference between the intervention and control groups)/the post-intervention level in the control group.
R E S U L T S
Description of studies
See: Characteristics of included studies; Characteristics of excludedstudies.The main literature search for all pharmaceutical policies resultedin 17,000 references to sift. The updated search for cap and co-payment policies in MEDLINE/EMBASE, Science Citation In-dex and web sites resulted in 4803 references. Including referencelists from relevant studies and reports, we sifted approximately22,000 references. We retrieved full text copies of 411 papers thatwere potentially relevant for the cap and co-payment group. Weexcluded 367 of these papers, most of them because they did notmeet the study design inclusion criteria. They were primarily re-views, editorials, modelling studies, cross-sectional studies and be-fore-after studies without a control group. The table ’Characteris-tics of excluded studies’ provides reasons for those studies whoseexclusion readers might question, studies that are well known butdo not meet all of inclusion criteria, and ITS studies that meet allinclusion criteria, except for an insufficient number of data points.We identified 11 studies (Almarsdottir 2000; Andersson 2006;Chang 2005; Dormuth 2006; Gibson 2005; Holloway 2001a;Holloway 2001b; Kephart 2007; Landsmann 2005; Lee 2006;Liu 2004) that possibly meet the inclusion criteria, but were notassessed before submission of the review. We have listed thesestudies in the ’Studies awaiting assessment’ section of the referencelist.Some of the studies reported on more than one intervention, andsome interventions were evaluated with more than one study de-sign. Overall 21 studies, reporting on 30 interventions, met theinclusion criteria.
Study designs
For some of the interventions, the effects were measured usingmore than one design, i.e. different designs for different outcomes.We included four interventions with RCT analyses (Newhouse1993); three interventions with RM analyses (Motheral 2001;Soumerai 1994; Tamblyn 2001); 12 interventions with ITS analy-ses (Blais 2002; Cromwell 1999; Donnelly 2000; Fryatt 1994; Hux1997; Martin 1996; McManus 1996; Ong 2003; Reeder 1985;Sawyer 1982) and 14 interventions with CBA analyses (Brian1974; Harris 1990; Huskamp 2005; Kozyrskyj 2001; Lingle 1987;Motheral 1999; Motheral 2001; Poirier 1998; Soumerai 1994).
For more information, please see the table, ’Characteristics of in-cluded studies’.
Characteristics of setting and interventions of
included studies
Overall, 20 interventions (in 13 studies) were from the USA (Brian1974; Cromwell 1999; Donnelly 2000; Harris 1990; Huskamp2005; Lingle 1987; Martin 1996; Motheral 1999; Motheral 2001;Newhouse 1993; Reeder 1985; Sawyer 1982; Soumerai 1994);five interventions (in five studies) were from Canada (Blais 2002;Hux 1997; Kozyrskyj 2001; Poirier 1998; Tamblyn 2001); two in-terventions (in one study) were from Australia (McManus 1996);one intervention (in one study) was from Nepal (Fryatt 1994) andtwo interventions (in one study) were from Sweden (Ong 2003).Eleven interventions (in nine studies) were fixed co-payments(Brian 1974; Fryatt 1994; Harris 1990; Hux 1997; Lingle 1987;Motheral 1999; Ong 2003; Reeder 1985; Sawyer 1982); six in-terventions (in three studies) were coinsurance with ceiling poli-cies (Blais 2002; Newhouse 1993; Tamblyn 2001); four interven-tions (in four studies) were cap policies (Cromwell 1999; Donnelly2000; Martin 1996; Soumerai 1994); three interventions (in twostudies) were fixed co-payment with ceiling policies (McManus1996; Poirier 1998); three interventions (in two studies) were tierco-payments (Huskamp 2005; Motheral 2001); one intervention(in one study) was a ceiling policy (Kozyrskyj 2001); one interven-tion (in one study) was a fixed co-payment and coinsurance witha ceiling policy (Ong 2003); and one intervention (in one study)was a fixed co-payment with cap policy (Harris 1990).See Additional Table 5 for further details.
Characteristics of outcomes
The studies provided data on drug use, i.e. the number of dis-pensed doses or the number of dispensed prescriptions (19 stud-ies); drug expenditures from a drug insurer’s perspective (10); drugexpenditures from the patient’s perspective (3); health care expen-ditures (2); overall drug expenditures (4); health outcomes (1) andhealth care utilisation (5). None of studies included interventioncosts (i.e. policy administration costs).
Risk of bias in included studies
Of the 30 included evaluations of interventions, seven had no lim-itations; three had no/some limitations (for different outcomes);16 had some limitations; one had some/serious limitations andthree had serious limitations.See Additional Table 3 and Table 4 for further details.
Effects of interventions
11Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
We attempted to order interventions according to their intensity,based on our assessment within each policy category of the size ofco-payment, which drugs were covered and the population thatwas included. Unfortunately, with the exception of cap policies,this was not possible since little information was reported on theintensities of policies. For policies other than caps, we have orderedinterventions chronologically. All payments are reported in theirlocal currency (see abbreviations Additional Table 6).
Caps
Four interventions met the inclusion criteria, of which three wereevaluated with ITS study (Cromwell 1999; Donnelly 2000; Harris1990; Martin 1996) and one with RM and CBA studies (Soumerai1994). We have ordered these studies based on our assessment ofintervention intensity, from most to least intensive.A summary of the findings for cap policies is provided in Addi-tional Table 7.1. Restricting reimbursement to three prescriptions
versus no restrictionsSetting: 1981, low-income patients (Medicaid) in New Hamp-shire, USA (Soumerai 1994)Study design: RM with no limitations and CBA with some limi-tationsThe introduction of this policy caused an immediate and sustaineddrop in drug use. The monthly drug use per person in the cohortof multi-drug users had a reduction of 46.0% (P<0.05), whilethe use by other patients was reduced by 17.0% (P < 0.05). Aloophole was embedded into this cap policy, allowing physiciansto triple the allowable quantity of pills and during the cap periodthe average prescription size increased by 13.0% (P < 0.05) in thecohort of multi-drug users.Overall, a large reduction of 58.0% (P < 0.05) in the monthlydrug use (per 100 eligibles) was found for “limited efficacy” drugs.However, there was also a reduction in “symptom relieving” drugs(38.0%, P < 0.05) and in “essential” drugs (28.0%, P < 0.05)(Soumerai 1994).
Reductions in drug use were also found in vulnerable sub-groups ofelderly patients (35.0%, P < 0.001), and severely disabled patientsdiagnosed with schizophrenia in the use of antipsychotic agents(15.4%, P < 0.003), anxiolytic and hypnotic agents (37.3%, P <0.001) and anti-depressants and lithium (49.1%, P < 0.001).Overall, the average plan reimbursements dropped by 38.0% (P <0.05); the effect declined and went close to pre-cap levels after thepolicy was discontinued at 12 months post the introduction. Plandrug expenditures decreased by 19.0% per patient per month (P< 0.05).There were no statistically significant effects on hospitalisations inany of the cohorts.In the cohort of schizophrenics, there was an increase in days ofadmissions to state psychiatric hospitals of 17.0% (P < 0.001) per
patient per month. Similarly, the number of visits per patient permonth to two community mental health centres increased abruptlyby 43.0% and 57.0% (P < 0.001) respectively, and was sustainedthroughout the period.In the cohort of elderly (>65), the relative risk of admissions tonursing homes was RR 1.8 (95% CI 1.2 to 2.6) compared to thecontrol group, and the probability of nursing home admissionswas particularly high for chronic multi-drug users RR 2.2 (95%CI 1.2 to 4.1). The nursing home stays were found to be longterm (Soumerai 1994).2. Restricting reimbursement to one anti-ulcer prescription
item with one refill reimbursed at a time
versus no restrictionsSetting: 1992, low-income population (Medicaid) in Florida, USA(Cromwell 1999)Study design: ITS with no limitationsThe policy also restricted high-dose prescription treatment of acutedisorders to 60 days.The reduction in the number of doses reimbursed was 42.7%(95% CI -50.1% to -35.4%). At one-year follow up, the reductionwas 39.6% (95% CI, -49.0% to -30.3%).Plan drug expenditures for anti-ulcers had an immediate reduction(dollars reimbursed) of 37.8% (95% CI -45.1% to -30.5%). Atone-year follow up, the reduction was 32.0% (95% CI -40.7% to-23.3%).There were not statistical significant changes in hospitalisationrates for any of the three conditions studied. The immediate effectswere 7.4% (95% CI -17.1% to 32.0%) for complicated PUD; -10.0% (95% CI -29.6% to 9.6%) for uncomplicated PUD; and15.6% (95% CI -9.9% to 41.0%) for non-ulcer peptic conditions.At one-year follow up, the effect was 9.3% (95% CI -24.7% to43.3%) for complicated PUD; 15.4% (95% CI -11.1% to 49.9%)for uncomplicated PUD; and 0.8% (95% CI -32.3% to 33.9%)for non-ulcer peptic conditions.3. Restricting reimbursement to five prescriptions
versus six prescriptions reimbursedSetting: 1991, low-income multi-drug users (Medicaid) in Geor-gia, USA (Martin 1996)Study design: ITS with some limitationsThe average monthly prescriptions per patient had an immediatereduction of 5.9% (95% CI -9.4% to -2.4%), at six months postthe reduction was 7.8% (95% CI -11.8% to -3.8%).Use of drugs paid for by the plan was reduced by 16.5% (95%CI -17.8% to -15.3%) per person, and at six months post thereduction was 17.7% (95% CI -19.1% to 16.3%). By comparison,the drugs paid for out of pocket by patients per month increasedby 26.5% (95% CI 16.5% to 36.5%), and at six months post wasincreased by 12.9% (95% CI -0.5% to 26.4%).4. Twenty days minimum re-supply period cap for drugs with
five or more repeats
versus three day minimum re-supply periodSetting: 1994, Pharmaceutical Benefits Scheme (all residents are
12Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
eligible), Australia (Donnelly 2000)Study design: ITS with no limitationsThis policy was introduced to manage the problem of a large sea-sonal effect (in December) after a previous introduction of a ceil-ing policy that had resulted in hoarding of drugs after patientshad reached the maximum co-payment level. Prior to the 20 daysre-supply cap (i.e. a cap on the time allowed until re-supply waspossible) policy, there were marked increases in utilisation towardsthe end of the year. After the introduction of the policy, the ab-solute number of prescriptions dispensed in December decreasedby 1,150,196 (95% CI 708,333 to 1,592,059) (estimated to beabout a 20% relative reduction). The effect of the policy appearedto be decreasing over time.
Fixed co-payments
Eleven interventions in 10 studies met the inclusion criteria, ofwhich five interventions used ITS designs (Fryatt 1994; Hux 1997;Ong 2003; Reeder 1985; Sawyer 1982) and six interventions hadCBA designs (Brian 1974; Harris 1990; Lingle 1987; Motheral1999). One intervention could not be included in the ’Results’section as the analysis was done inappropriately and data not re-analysed (Fryatt 1994).A summary of the findings for fixed co-payment policies is pro-vided in Additional Table 8.1. Fixed co-payment increases in tier co-payment systems
(Motheral 1999)a. USD 7 fixed co-payment per generic & USD 15 per brand drugversus USD 5 generics and USD 10 for brand drugsSetting: 1997, commercial plan, USAStudy design: CBA with serious limitationsOne commercial plan in the USA increased the co-payments perprescription for generics from USD 5 to USD 7, and brand drugsfrom USD 10 to USD 15. The experimental commercial plan wascompared to a control with a USD 5 for generics and USD 10 forbrand drugs. No P-values or confidence intervals were reported.Overall drug use (claims per patient) was reduced by 22.5%, andthe mean generic fill rate per patient increased by 30.0%.The mean plan expenditures per patient decreased by 28.8%, whilethe mean patient co-payment per patient increased by 32.2%.The mean ingredient expenditures per patient decreased by 17.3%,and the mean expenditure per claim per patient was reduced by4.7%.b. USD 5 fixed co-payment per generic & USD 15 per brand drugversus USD 4 for generics and USD 10 for brand drugsSetting: 1997, commercial plan, USAStudy design: CBA with some limitationsOne commercial plan in the USA increased the co-payments perprescription for generics from USD 4 to USD 5, and brand drugsfrom USD 10 to USD 15. The experimental commercial plan wascompared to a control with a USD 5 copayment for generics andUSD 10 for brand drugs. No P-values or confidence intervals were
reported.Overall drug use (claims per patient) was reduced by 21.3%, andthe mean generic fill rate per patient increased by 23.3%.The mean plan expenditures per patient decreased by 33.3%, whilethe mean patient co-payment per patient increased by 39.8%The mean ingredient expenditures per patient decreased by 19.4%,and the mean expenditures per claim per patient were reduced by3.5%.2. Income based: CAD 100 full co-payment, after which the
patients had to pay CAD 6.11 or CAD 2 (low-income) per
prescription
versus full drug coverageSetting: 1996, Ontario drug benefit program (all seniors eligible),Canada (Hux 1997)Study design: ITS with some limitationsFor all classes of drugs studied, the number of prescriptions de-creased, but the quantity per prescription increased. The overallreduction in number of prescriptions was 14.2%, and volume wasreduced by 6.0% at five months post policy.For “essential drugs”, there was a reduction in the number of pre-scriptions ranging from 10.3% to 15.9% (P not reported). Theeffects on volume were reported to be statistically not signifi-cant, ranging from -1.3% to +2.6%.with the exception of antipsy-chotics, which had a reduction of 8.7% (P < 0.05).For all “discretionary drugs”, there was a reduction in the numberof prescriptions, ranging from 14.3% to 24.3% (P not reported).The reduction in volume for most discretionary drugs was statis-tically significant (P < 0.05) and ranged from -10.9% to -20.0%,with the exception of sedatives (-4.0%, P > 0.05).The use of lipid-lowering (“preventive”) drugs had an 11.3% re-duction in the number of prescriptions (P not reported) and 1.7%in volume (P not reported).The plan drug expenditures decreased by 16.9% (P not reported).3. SEK 160 initial fixed co-payment after which patients paid
SEK 60 for additional drugs versus SEK 125 initial fixed co-
payment and SEK 25 for additional drugs
Setting: 1995, public health insurance system (all eligible), Sweden(Ong 2003)Study design: ITS with no limitationsThe effects of the policy were examined separately for men andwomen, since the prevalence of depression, anxiety and sleep dis-orders in Sweden was found to be higher for women. Women’suse of drugs (Defined daily doses (DDD) per 1000 patients) tem-porarily increased immediately preceding the policy by 6618 (P< 0.01) for antidepressants, 2861 (P < 0.01) for anxiolytics and8734 (P < 0.01) for sedatives. However, there were no statisticallysignificant sustained changes in the use of any of the drugs postpolicy (figures not reported).Men’s use (DDDs per 1000 patients) of all three drugs increasedin the two months before the policy by 2749 (P < 0.01) for an-tidepressants, 3477 (P < 0.01) for sedatives and 1759 (P < 0.01)for anxiolytics. This trend was interrupted for the use of antide-
13Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
pressants (-5275, P < 0.01) and sedatives (-5838, P < 0.01), butcontinued to rise after the policy. There was not a statistically sig-nificant change post policy in prescriptions for anxiolytics (figuresnot reported).4. Fixed co-payment increased in two phases (Harris 1990)a. USD 3 fixed co-payment per prescriptionversus full drug coverageSetting: 1984, a group health non-profit staff HMO, USA (Harris1990)Study design: CBA with some limitationsOverall items per patient decreased by 10.6% (P < 0.0001), “es-sential” drugs had a reduction of 13.0% (P < 0.0001) and “discre-tionary” drugs of 19.2% (P < 0.0001).Plan drug expenditures per prescription increased by 10.3% (P< 0.01), and average drug expenditures per patient decreased by5.2% (P < 0.01).b. USD 1.5 fixed co-payment per prescriptionversus full drug coverageSetting: 1983, a group health, non-profit, staff HMO, USAStudy design: CBA with some limitationsOverall items per patient decreased by 10.7% (P < 0.001); “es-sential” drugs had a reduction of 10.5% (P > 0.05), and “discre-tionary” drugs of 17.3% (P < 0.001).Plan drug expenditures per prescription increased by 4.4% (P <0.01), and average drug expenditures per patient decreased by6.7% (P < 0.001).5. USD 0.50 fixed co-payment per prescription
versus full drug coverageSetting: 1977, low-income patients (Medicaid) South Carolina,USA (Reeder 1985)Study design: ITS with no/some limitationsOne study measured the effect of a USD .50 co-payment perprescription in a Medicaid population (Reeder 1985). Results werereported as absolute changes in level per person.The overall use was reduced by 0.3 prescriptions per person (P< 0.05), estimated to be about 12.0% below predicted values.The effect on individual drug groups ranged from +0.1 to -0.4prescriptions per person.Drug expenditures per person decreased by USD 2 per person permonth (P < 0.05), estimated to be 16.0% below predicted values.6. USD 0.50 fixed co-payment per prescription
versus full drug coverage including most over-the-counter drugsSetting: 1976, low-income elderly patients (Medicaid) in Mary-land, USA (Sawyer 1982)Study design: ITS with some limitationsThe policy also included restrictions on reimbursement for over-the-counter drugs.The policy had a no statistical significant change of 0.1% (95%CI -15.0% to 15.2%) on the overall plan drug expenditures.7. USD 2 fixed co-payment per prescription for elderly
versus no drug coverageSetting: 1975, elderly low-income patients (Medicare) in New
Jersey/Pennsylvania, USA (Lingle 1987)Study design: CBA with some limitationsData on the use of healthcare services and expenditures were notreported when the results were not statistically significant. Resultsare reported here (when available) as differences between groupsadjusted for baseline, or as differences between groups post policy.For most healthcare services, there was no statistically significantchange in reimbursements. An increase was found for surgeon ser-vices of USD 4.4 (P < 0.05) and for MD hospital services of USD2.5 (unadjusted from baseline) per patient (P < 0.05). Reductionswere found for supplier reimbursement expenditures of USD 54.1(unadjusted from baseline, P < 0.01), and for inpatient hospitalservices of USD 238.5 per patient (unadjusted from baseline, P <0.01) compared to the control.There were no statistically significant effects on the use of nurs-ing home facilities or inpatient services (hospital admissions) (P> 0.05). Some reductions were observed in the use of outpatientservices, non-hospital based health care, physician and surgeonservices in the experimental group versus the control (range -0.1to -2.8, P < 0.01), however for these outcomes (except for sur-geon payments) there were also statistically significant differencesin baseline between the groups, where utilisation was lower in theexperimental group. An increase was found in the average utilisa-tion per patient for MD hospital services of 1.0 (P < 0.05) com-pared to control.The relative risk of not dying was 1.2 at two years post policy inthe experimental group versus the control (95% CI 0.95 to 1.42).8. USD 0.50 per prescription
versus full drug coverageSetting: 1972, low-income patients (Medicaid) California, USA(Brian 1974)Study design: CBA with serious limitationsThe policy is also referred to as “The California Experiment”, andincluded an increase of USD 1 per each of the first two visits to ahealth provider. No P-values or confidence intervals were reportedin the studies.Drug use by families with dependent children was reduced by9.0% for critical drugs, 17.0% for needed drugs, 11.1% for broad-range drugs, 7.2% for preventive drugs and 8.6% for all otherdrugs at three months post policy.Drug use by the elderly was reduced by 8.7% for critical drugs,9.8% for needed drugs, 4.9% for broad-range drugs and 7.6% forall other drugs at three months post policy.Drug use by disabled was reduced by 5.2% for critical drugs, 6.2%for needed drugs, 6.2% for broad-range drugs, 0.3% for preventivedrugs and 7.8% for all other drugs at three months post policy.Fixed co-payment with capOne evaluation measured the effect of introducing a fixed co-payment with a cap with a CBA design (Harris 1990).USD 3 fixed co-payment per prescription for a 30-day supply
versus full drug coverage/no restrictionsSetting: 1985, a group health, non-profit, staff HMO, USA (
14Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Harris 1990)Study design: CBA with some limitationsThe policy also restricted reimbursement for over-the-counterdrugs and introduced a USD 5 fee for outpatient visits in July1985 (Harris 1990).Overall items per patient decreased by 12.0% (P < 0.001), “es-sential” drugs by 4.0% (P > 0.05), and “discretionary” drugs by19.0% (P < 0.001).Over-the-counter drug use decreased by 25.3% (P not reported).Drug expenditures per prescription increased by 8.5% (P < 0.01),and average drug expenditures per patient decreased by 8.8% (P< 0.001).
Ceiling
One study with CBA design reported on a ceiling policy (Kozyrskyj2001).Yearly income based ceiling of 2% (low-income) or 3% (high-
income)
versus full drug coverageSetting: 1996, children in Manitoba Pharmacare, Canada (Kozyrskyj 2001)Study design: CBA with some/serious limitationsBoth intervention groups (low- and high-income) had 40% coin-surance up to a ceiling of CAD 237 as policy in baseline. The over-all likelihood (ratio of odds ratios) of use of inhaled corticosteroidsfor children (per patient) with stable mild to moderate asthma was0.85 (CI 95% 0.74 to 0.96) at one year post and 0.78 (CI 95%0.68 to 0.88) at two-year follow up.The overall likelihood (ratio of odds ratios) of use of inhaled corti-costeroids for children (per patient) with stable severe asthma was0.81 (CI 95% 0.66 to 0.95) at one year post, and 0.82 (CI 95%0.68 to 0.96) at two-year follow up.Fixed co-payments with ceilingsThree interventions in two studies met the inclusion criteria, ofwhich two evaluations had ITS designs (McManus 1996) and onehad a CBA design (Poirier 1998).A summary of the findings for fixed co-payments with ceilings isprovided in Additional Table 9.1. Income-based fixed co-payment with ceiling (McManus1996)a. AUD 15 (community) or AUD 2.50 (elderly) fixed co-payment perprescription with ceilingversus AUD 11 fixed co-payment per prescription with ceiling(community), full drug coverage (elderly)Setting: 1992, community and elderly patients, PharmaceuticalBenefits Scheme, AustraliaStudy design: ITS with some limitationsThe level of the ceiling was not reported.The use of “essential” drugs decreased by 816,000 prescriptions (P< 0.001, 95% CI; -1,116,133, -516,373), estimated to be 22.0%below the predicted level.
The use of “discretionary” drugs decreased by 758,500 prescrip-tions (P < 0.001, 95% CI; -901,189, -615,813), estimated to be27.0% below predicted levels.b. AUD 2.5 fixed co-payment per prescription with ceilingversus full drug coverageSetting: 1992, repatriation patients, Pharmaceutical BenefitsScheme in AustraliaStudy design: ITS with no limitationsThe level of the ceiling was not reported.One study measured the effects of increasing a fixed co-paymentfrom no co-payment to AUD 2.5 per prescription for repatriationpatients belonging to the Australian PBS scheme covering approx-imately 90% of the community (McManus 1996).The use of “essential” drugs decreased by 29,500 prescriptions (P< 0.001, 95% CI -45,812 to -13,287), estimated to be 23.0%below predicted levels.The use of “discretionary” drugs decreased by 32,500 (P < 0.001,95% CI; -44,442, -20,510), estimated to be 24.0% below thepredicted level.2. CAD 2 fixed co-payment per prescription up to a ceiling of
CAD 100
versus full drug coverageSetting: 1992, elderly patients in Quebec public health insurancedrug program (all eligibles), Canada (Poirier 1998)Study design: CBA with some limitationsThe use of antihypertensives was reduced by 2.3% (P < 0.05) forlow-income patients and 3.7% (P < 0.05) for high-income at 18months’ follow up.The use of benzodiazepines was reduced by 1.2% (P < 0.05) forlow-income patients and 1.3% for high-income (P < 0.05) at 18months’ follow up.
Coinsurance with ceiling
Six interventions in three studies met the inclusion criteria, ofwhich two interventions were analysed with ITS or RM de-signs (Blais 2002; Tamblyn 2001) and four with a RCT design(Newhouse 1993).A summary of the findings for coinsurance with ceiling policies isprovided in Additional Table 10.1. Income based coinsurance and ceiling (Blais 2002; Tamblyn2001)a. 25% coinsurance up to a ceiling of CAD 200versus full drug coverageSetting: 1996, low-income elderly and welfare patients, Quebecpublic health insurance drug program (all eligibles), CanadaStudy design: RM/ITS with no limitationsThe overall use of “essential” drugs declined by 17.7% (95% CI-14.8% to -20.5%). Within this group, anti-convulsants had animmediate reduction of 16.2% (95% CI -9.0% to-23.4%); inhaled corticosteroids of 55.6% (95% CI -49.8% to -64.4%); and neuroleptics by 15.5% (95% CI -9.9% to -21.8%).
15Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
The overall use of “less essential” drugs had a reduction of 19.4%(95% CI -17.4% to -21.4%).b. 25% coinsurance up to an income based ceiling of CAD 200, CAD500 or CAD 750versus CAD 2 fixed co-payment per prescription, up to a ceilingof CAD 100Setting: 1996, elderly patients, Quebec public health insurancedrug program (all eligibles), CanadaStudy design: RM/ITS with no limitationsThe overall use of “essential” drugs declined by 6.9% (95% CI -5.5% to -8.4%).Within this group, antihypertensives had an immediate reductionof 16.9% (95% CI -12.0% to -21.9%); anti-coagulants of 17.2%(95% CI -12.7% to -21.7%); and nitrates by 22.6% (95% CI -17.6% to -27.7%).The overall use of “less essential” drugs experienced a reductionof 14.0% (95% CI -13.0% to -15.0%). The effect on benzodi-azepines was measured and utilisation was found to decline by23.4 (95% CI -17.1% to -29.6%).2. Coinsurance on both drugs and services with an income-
based ceiling (Newhouse 1993)a. 95% coinsurance up to an income-based ceiling of 5%, 10% or15%, or maximum USD 1000versus full drug and service coverageSetting: 1977, patients from six areas of USA (the RAND healthinsurance experiment)Study design: RCT with some limitationsOverall use of prescription drugs was reduced by 33.6% (P < 0.05),and the overall use of over-the-counter drugs by 33.5% (P = 0.05)as compared to control.Overall prescription drug expenditures decreased by 37.6% (P <0.05), and over-the counter drug expenditures by 35.0% (P >0.05).b. 50% coinsurance up to an income-based ceiling of 5%, 10% or15%, or maximum USD 1000versus full drug and service coverageSetting: 1977, patients from six areas of USA (the RAND healthinsurance experiment)Study design: RCT with some limitationsOverall use of prescription drugs was reduced by 23.2% (P < 0.05),and over-the-counter drugs by 59.8% (P < 0.05), as compared tocontrol.Overall prescription drug expenditures decreased by 33.6% (P< 0.05), and over-the-counter drug expenditures by 58.9% (P <0.05).c. 25% coinsurance up to an income-based ceiling of 5%, 10% or15%, or maximum USD 1000versus full drug and service coverageSetting: 1977, patients from six areas of USA (the RAND healthinsurance experiment)Study design: RCT with some limitationsOverall use of prescription drugs had a reduction of 18.4% (P <
0.05), and overall use of over-the-counter drugs was reduced by15.5% (P > 0.05), as compared to control.Overall prescription drug expenditures decreased by 8.3% (P >0.05), and over-the-counter drug expenditures by 26.8% (P >0.05).d. 95% coinsurance up to a ceiling of USD 150 per person or USD450 per familyversus full drug and service coverageSetting: 1977, patients from six areas of USA (the RAND healthinsurance experiment)Study design: RCT with some limitationsInpatient services in this policy were free, but outpatient serviceswere not covered.Overall use of prescription drugs was reduced by 18.6% (P < 0.05)and over-the-counter drugs by 5.9% (P > 0.05).Overall prescription drug expenditures decreased by 16.3% (P >0.05) and over-the-counter drug expenditures by 6.1% (P > 0.05).
Fixed co-payments and coinsurance with a ceiling
One study reported the effects of a fixed co-payment with coin-surance and a ceiling with an ITS design (Ong 2003). The effectsof the policy were examined separately for men and women, sincethe prevalence of depression, anxiety and sleep disorders in Swe-den was found to be higher for women.SEK 400 initial prescription payment, after which patients paida proportion of the additional expenditures up to an annual co-payment ceiling of SEK 1300versus SEK 160 initial prescription payment, and SEK 60 foradditional drugsSetting: 1997, public health insurance system (all eligible), Sweden(Ong 2003)Study design: ITS with no limitationsFor women, there were immediate increases in the month pre-ceding the policy in the use (DDDs per 100 patients) of antide-pressants (16,095) (P < 0.01), anxiolytics (2739) (P < 0.01) andsedatives (12,201) (P < 0.01). After the introduction of the pol-icy, antidepressants had a sustained change in use of -21,129 (P< 0.01). The use of anxiolytics had a change of -3548 (P < 0.01)that decayed over time in the 2.5-year follow up. Sedatives had animmediate and not sustained change of -11,304.There were no sustained effects in the use of drugs by men. Aswith the women’s cohort, there were increases in drug use (DDDsper 100 patients) immediately preceding the policy; antidepres-sants and sedatives changed by 10,474 (P < 0.01) and 7703 (P <0.01) respectively, whereas anxiolytics had no statistically signifi-cant change pre-policy. Post policy there were changes in the useof all three drugs, but the effects decayed over time (-4.393 (P <0.01) for antidepressants, -1600 (P < 0.01) for anxiolytics and -3415 for sedatives (P < 0.01)).
16Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Three interventions in two studies reported on the effects of tierco-payments (Huskamp 2005; Motheral 2001), of which two in-terventions were evaluated using CBA designs (Huskamp 2005),and one intervention by a mixed CRM/CBA design (Motheral2001).A summary of findings for tier co-payments is provided in Addi-tional Table 11.1. Change in tier co-payment systems (Huskamp 2005)a. three-tier plan and increased fixed co-payments by USD 1 pergeneric drug (1991, USA)versus two-tier plan (one-tier plan in baseline)Setting: 1991, large employer health plan, USAStudy design: CBA with some limitationsAll results are given as unadjusted differences of differences, oneyear post policy. Long-term effects were not available.The use of ACE inhibitors and statins across all tiers decreasedby 24.0% (P < 0.001) and 24.0% (P < 0.001) respectively. ForPPIs, brand drugs available in tiers two and three, the reductionwas 34.0% (P < 0.001). In a vulnerable cohort of children usingADHD medications (available all tiers), there was a 17.0% reduc-tion in the probability of use (P < 0.001).For ACE inhibitors and statins, the reductions in overall drug ex-penditures across tiers were 0.3% (P = 0.59) and 0.7% (P = 0.301)respectively. The overall drug expenditures for PPI brand drugsacross tiers two and three were reduced by 3.2% (P < 0.001). In thepopulation of children using ADHD medications, the reductionwas 3.0% (P = 0.23).The plan drug expenditures were reduced by 58.2% (P < 0.001)for ACE inhibitors and 13.7% (P < 0.001) for statins. Plan drugexpenditures for PPI brand drugs were reduced by 15.3% (P <0.001). A decrease could also be seen in monthly plan spendingon ADHD medications for children, with a reduction of 43.0%(P < 0.001).Patient expenditures increased by 141.8% (P < 0.001) for ACEinhibitors and 117.9% (P < 0.001) for statins. The largest increasein patient co-payment could be found for PPI brand drugs, whichincreased by 148.0%. Patient drug expenditures for children usingADHD medications increased by 46.0% (P < 0.001).b. three-tier planversus two-tier planSetting: 1991, large employer health plan, USA (Huskamp 2005)Study design: ITS with no/some limitationsThe co-payments in baseline were USD 6 for tier one drugs, USD12 for tier two drugs and USD 24 for tier three drugs. The compar-ison was matched to experimental group’s baseline co-paymentsfor tier one and tier two. All results are given as unadjusted differ-ences of differences, one year post policy. Long-term effects werenot available.The overall use of ACE inhibitors and statins across tiers wasreduced by 5.0% (P = 0.26) and 2.0% (P = 0.69) respectively. ForPPIs, (available in tiers two and three), the reduction was 5.0% (P
= 0.32).Drug expenditures for ACE inhibitors and statins increased acrossall tiers by 3.1% (P < 0.001) and 2.0% (P < 0.02) respectively. PPIbrand drugs had a reduction of 0.4% (P = 0.66).The change in plan drug expenditures showed mixed results. ACEinhibitor expenditures were reduced by 5.6% (P < 0.001), statinsincreased by 1.9% (P = 0.07), and PPI brand drugs were reducedby 2.3% (P < 0.02).Patient expenditures increased by 7.5% (P < 0.001) for ACE in-hibitors and by 0.3% for statins (P = 0.075). Patient expendituresfor PPI brand drugs increased by 4.9% (P < 0.001).2. three-tier plan and increased fixed co-payment by USD 1
per generic drug (1998, USA)
versus two-tier planSetting: 1998, large commercial health plan, USA (Motheral2001)Study design: RM/CBA with some limitationsOverall drug use across tiers was reduced by 5.4% at one-yearfollow up (P < 0.001). There was no statistically significant changeat two-year follow up.The reduction in preferred tier two drugs and non-preferred tierthree drugs was 3.8% (P < 0.003) and 21.8% (P < 0.001), respec-tively, at one year post. After two years, preferred tier two branddrugs were back at 0.6% over baseline levels, but the reduction inuse of non-preferred tier three brand drugs was sustained at 23.8%(P < 0.01). Use of generics (one tier) was reduced by 2.7% (P >0.05).In the category consisting of drugs not classified in any of the tiers,the reduction in drug use was 11.6% (P < 0.004) at one year post,and at two year post the reduction was 10.4% (P < 0.05) (Motheral2001). This group of drugs consisted of drugs that changed tierstatus or drugs with no substitution possibilities.Overall drug expenditures were reduced by 3.2 (P < 0.001) meanmonthly claim expenditures per patient, estimated to be 12.2%below predicted level. Plan drug expenditures were reduced by4.8% (P < 0.001) mean monthly plan net expenditures per patientone year post, estimated to be 25.5% below predicted level. Patientdrug expenditures increased by 1.6 (P < 0.001) mean monthlyclaim expenditures per patient, estimated to be 22.8% over pre-dicted level. The effects were sustained throughout the long-termfollow up.Changes in healthcare utilisation were not statistically significant(P > 0.05). Physician visits increased by 0.0010 (estimated to be0.4% over predicted level), hospitalisations by 0.0007 (estimatedto be 26.9% over predicted level), and ER visits per patient permonth by 0.0011 (estimated to be 8.7% over predicted level).
D I S C U S S I O N
17Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Restricting reimbursement using a cap reduced drug use and savedplan drug expenditures. It also had the unintended effect of reduc-ing necessary drug use when applied to “essential” drugs, and putextra strain on already vulnerable populations. In one study thisresulted in an increased use of healthcare services and deteriorationof health in these populations (Soumerai 1987). Few evaluationswere included and the overall quality of the evidence was low.
The largest reductions in drug use were found in the policies thatwe judged to be most intensive, restricting reimbursement to threedrugs for all drugs (including “essential” drugs) per month inNew Hampshire (Soumerai 1994) and restricting reimbursementto only one anti-ulcer drug with one refill in Florida (Cromwell1999).Cap policies were expected to be most intensive for multi-drugusers; this was supported by one study that included outcomeson both multi-drug users and other drug users (Soumerai 1994).Normally, multi-drug users are considered to be less likely to dis-continue their drug use, since they are more likely to be sick andthus more dependent on their medications. However, the cohortsof multi-drug users in this study were low-income and thus thepatients may have been particularly sensitive to the restriction. In-terestingly, the policy that restricted reimbursement from six tofive prescriptions in Georgia only had a modest effect on druguse, despite the study cohort only including low-income, multi-drug users. One reason may be that since a cap on six drugs al-ready was implemented, the patients may have been less sensitiveto the restriction. The increase in patient out-of-pocket expensesindicates that many (approximately 75% of the patients) choseto continue their drug use and to cover the expenses themselves.The investigators also emphasised that the estimated effect may beconservative since the drugs put on prior approval were excludedfrom the analysis and that, when a recipient was affected by bothpolicies (i.e. a multi-drug user with six prescriptions per monthincluding benzodiazepines), the patient may have been more in-clined to drop the drug that was under prior approval (Martin1996).Introducing a restriction on a certain number of prescriptions ordrugs is also intended to stimulate the patient to prioritise the useof the most important drugs. However, both “essential” and otherdrugs had substantial reductions, so whether the patients failedto prioritise their drug use, or the number of needed “essential”drugs exceeded the cap limit, or other factors were involved, isunknown.In New Hampshire, the largest reduction in the actual numberof prescriptions was found for several commonly used “essential”drugs (insulin -28%; furosemide -30%; thiazides diuretics -28%and digoxin -45%).A cap on prescriptions or prescription items, but not on volume,may offer a loophole for physicians and consumers to increasethe doses prescribed per prescription. In New Hampshire this was
possible, and although authors found that the volume prescribedper prescription within the multi-drug users’ cohort increased by13% (P < 0.05), it did not offset the effect of the policy (Soumerai1994).No direct data on health effects were reported in the studies, sonone could be included in the analysis; however Soumerai et al(Soumerai 1994) state that there was “no increase in mortality” inNew Hampshire.There were no statistically significant changes found in hospitali-sations in studies by Cromwell et al (Cromwell 1999) (Medicaidusers) and Soumerai et al (Soumerai 1994) for elderly Medicaidmulti-drug users. Nevertheless, the changes in the use of otherhealthcare services in the study by Soumerai et al (Soumerai 1994)for vulnerable patients are worrying.Introducing a cap in New Hampshire caused an increase in nursinghome admissions for elderly people, and an increase in days ofadmissions to state psychiatric hospitals and visits to the mentalhealth centers for schizophrenics (Soumerai 1994). It should benoted that since nursing home residents were exempted from theprescription cap, it is difficult to tell if the rise in nursing homeadmissions was caused by a deterioration of health or to avoid theeconomic barrier.Soumerai et al (Soumerai 1994) reports that the savings in drugreimbursements were estimated to be about USD 780,000 peryear for the cohort of 34,922 patients (Feb 1982 USD), but em-phasised that even though the policy may have saved plan drug ex-penditures, the savings may be low compared to the potential sideeffects from “essential” drug discontinuation and the increased useof healthcare services in certain vulnerable populations (Soumerai1994). This was certainly the case in the cohort of severely disabledschizophrenics, where the expenditures of mental health servicesincreased by USD 390,000 in the study period (Soumerai 1994).The last of the cap policies included in this review measured theeffect on restricting the supply period in Australia. The intendedeffect behind this policy was somewhat different from the policiesdiscussed above, and was thus not included in the ’Summary offindings’ table. In an attempt to manage the stockpiling effectinduced by implementing a maximum co-payment ceiling policythe year before, the cap on supply period attempted to reducethe hoarding of drugs and to distribute the supply of drugs moreevenly throughout the year. After the introduction of this policy,there was a statistically significant decrease in the peak utilisationin December. Although the effect appeared to decrease over time,it was still under pre-policy levels, even five years after the policywas introduced. The policy had no statistically significant changeon the January levels (compared to baseline) as one would suspectbecause, during January, patients had not yet reached their co-payment ceiling and thus the cap policy would not have had anyeffect on utilisation in this month.
Fixed co-payments
18Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Introducing fixed co-payment reduced drug use across studies,even though the co-payments required by the patients in mostcases were quite small. Use of both “essential” and other drugs wasreduced, although the reduction in “essential” drugs was somewhatless. Substantial reductions in drug plan expenditures were found,whereas there were only modest reductions in the overall drugexpenditures, suggesting a shift of cost from plan to patients.None of the included evaluations reported effects on health orhealthcare utilisations. Few evaluations were included and thequality of the evidence was moderate to very low.Fixed co-payments are intended to reduce drug use and partiallyshift costs from the insurer to patients. Generally, the evidence ofthis review supports this, although it is difficult to estimate thelevel of the policies’ intensity or magnitude of effect across studies.The reason for this was the heterogeneity of studies, and that littleinformation was reported on relative increase in the co-paymentlevel, price year and participants’ “disposable income” relevant tothe level of co-payment.It is also difficult to detect any apparent differences in utilisationacross patient groups based on income or other variables. Onestudy evaluated the effects of an income-based policy introducedin Canada, but the effects were reported aggregated for the lowand higher income groups (Hux 1997). This is unfortunate, sincenot only may the patients have reacted differently to a co-paymentpolicy, but they also received interventions of different intensity.The study by Brian et al (Brian 1974) found some differential ef-fects between the included study cohorts. Socio-economic differ-ences were identified between the aid groups; for example disabledand elderly were more often middle-income, while this was notthe case for the families with dependent children that also hadthe greatest reduction in drug use. The two commercial insuranceplans in the USA that increased the differential co-payments onbrand drugs and generics experienced an overall reduction in druguse and an increased use of generics. This suggests that, despiteincreased co-payments on generics, patients chose to switch fromthe higher co-payments for brand drugs (Motheral 1999).Overall, the reductions in drug use and in plan drug expendituresmay have led to cost savings for the insurers. Hux et al (Hux 1997)estimated the savings in Ontario, Canada to be about CAD 13million per month (price year not reported). However, since effectson health, healthcare utilisation and patients’ expenditures werenot reported, we do not know what effects the discontinuation ofdrugs may have had on, for example, vulnerable populations orhealthcare utilisation.Hux et al (Hux 1997) compared the decrease in antipsychoticsfound in Manitoba, Canada to the reduction that was found inNew Hampshire after the introduction of a cap policy (Soumerai1994). As discussed above, the discontinuation of these drugs ledto higher rates of ER use and use of other health services in this pa-tient cohort. The authors note that this group seem to be particu-larly vulnerable to the adverse effects of co-payments (Hux 1997).Similarly, Reeder et al (Reeder 1985) found substantial reductions
in psychotherapeutics and other “essential” or important drugs,such as diuretics and cardiovascular drugs, which the authors noteare not likely to be over-prescribed.
In some of the interventions, different factors may have played apart in modifying the expected effect. The study of the Canadianintervention (Hux 1997), which perhaps was the most intensive ofthe policies included (CAD 100 initial payment, after which pa-tients paid CAD 6.11), reported that the physicians, in an attemptto protect their patients from the co-payment, may have changedtheir prescribing patterns. An increase in volume per prescriptionfor “essential” drugs nearly offset the drop in number of prescrip-tions. This was not the case for “discretionary” drugs, where thereduction in quantity had a much larger and statistically signif-icant reduction (P < 0.05). Similarly, the study from Maryland,where the introduction of a USD 0.50 fixed co-payment per pre-scription and elimination of coverage for over-the-counter drugswas expected to decrease plan expenditures, found no statisticallysignificant effects on plan drug expenditures (Sawyer 1982). Al-though there was a decline in drug use, the average expenditureper prescription had increased. The reason for this was that intro-duction of the patient co-payment led physicians to increase thesize of the prescriptions as a way of lowering the expenditures forpatients (Sawyer 1982).Physicians and pharmacists may also have counteracted the USD1.5 policy introduced in California (Brian 1974). Although theywould have had to pay the co-payment themselves, physicians andpharmacists may have exempted patients from the co-payment.The impact of this is unclear. For example, the investigators (Brian 1974) reported that for families of dependent children, only82.9% had paid the USD 0.50 drug co-payment. Similarly, thismay have been the case for the USD 0.50 policy in South Carolina,but the investigators reported that the impact in this case was small(Reeder 1985).Stockpiling was another factor in the Canadian study (Hux 1997).The abrupt and marked reduction in number of prescriptions thattook place immediately after the policy may have been due to thestockpiling of drugs that took place prior to implementation ofthe co-payment policy.The California co-payment experiment of 1972 (Brian 1974) hadmethodological limitations that may have underestimated the ef-fects. The co-payment group had a somewhat higher income thanthe control group. There was also a problem with collecting all theclaims, which may have yielded lower than actual utilisation rates.In one of the two evaluations of differential co-payments reportedby Motheral et al (Motheral 1999), the intervention and controlgroups had different policies at baseline. While the interventiongroup started at USD 4 for generics and increased to USD 5with the implementation of the new policy, the control group hada USD 5 co-payment for generics throughout the study period.Therefore the reported effects of this increase in co-payments forgenerics may underestimate the effects of this policy change.
19Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Three studies could not be included in the ’Summary of findings’table. The study by Ong et al (Ong 2003) only reported absolutefigures, and statistically non-significant results were not reported.The policy had no statistically significant effects post policy, andonly temporary reductions immediately post policy for some druggroups. The only sustained effects were for men’s use of antidepres-sants and sedatives (Ong 2003). The policy was introduced to thegeneral public in Sweden, and was one of two policies succeedingeach other. Ong et al (Ong 2003) explained the modest effect bysuggesting that patients may have valued the drugs more highlythan the burden of increased co-payments. They also suggestedthat the increased utilisation by men may have reflected previousunder-treatment (Ong 2003).One study measured the effects of introducing co-payment in apopulation with no drug coverage in New Jersey and Pennsylva-nia (Lingle 1987). The introduction of partial drug coverage washypothesised to be more expensive for the plan, but this was dis-missed by the investigators based on the evidence of this study,since the expenditures of the outpatient drug program were offsetby the reduction in reimbursements for inpatient services (exclud-ing administration expenditures). The reduction in use of somehealthcare services may also indicate an improvement in health asa result of improved access to care; however this was not addressedby the investigators.Another study we were not able to include in the ’Summary of find-ings’ analysis because of methodological limitations was a studyfrom Nepal evaluating the effect of a fixed co-payment per pre-scription item (i.e fee per drug included in a prescription) versus afixed co-payment per prescription (Fryatt 1994). The study is theonly intervention from a low- or middle-income country that metour inclusion criteria. The policy was introduced for two reasons;to partly fund the “essential” drug supply needed at the healthposts, and to encourage more rational prescribing. The health postsare primary care units serving the general population in the EastNepalese hill districts, usually run by staff with a variable amountof health training (1-2 years). The health post included in thisstudy was supported by the Britain Nepal Medical trust.The already existing “fee per prescription” scheme was criticisedfor encouraging over-prescribing (Fryatt 1994). Patients attendingthe “fee per item” health posts had fewer items prescribed thanthe patients in the “fee per prescription” scheme (Fryatt 1994). Asfor the availability of “essential” drugs, the investigators reportedthat there was a larger proportion of drugs that was either “low” or“empty” in stock in the fee per prescription scheme (Fryatt 1994).Overall drug expenditures were less in the “fee per item” schemethan in the “fee per prescription” scheme. This was explained byfewer prescribed drugs and that the “fee per prescription” schemeprescribed more expensive drugs. The “fee per prescription item”scheme was associated with an increase in daily attendance com-pared to baseline. The authors explain this by suggesting that thecommunity may have perceived the new scheme to provide higherquality care because of the regular supply of drugs. The overall
drug subsidy needed at the health posts per patient in the fee peritem scheme had a reduction of 24%; however, the increase inattendance caused a higher subsidy for running the health postsof about USD 137 per month compared to USD 78 (June 1991USD) in the fee per prescription scheme.The study had several methodological limitations. The time seriesanalysis was not done appropriately and it was not possible toreanalyse the data. Moreover, there were large differences betweenthe health posts in the comparison groups, and strong patternsof seasonality. It is also difficult to know how comparable thepopulations in the intervention and control groups were.
Fixed co-payments with caps
Introducing a fixed co-payment policy with a cap and restrictionson over-the-counter drugs reduced all drug use and plan drugexpenditures in a non-profit, staff HMO in the USA (Harris1990). No effects on health or healthcare services were reported.The quality of the evidence was low.The policy was the third of three policies introduced in westernWashington succeeding each other. The three co-payment inter-ventions were compared to the same control group over time andadjusted with the data for the (baseline) year before. Comparingthe experimental groups and control in a time perspective, thedifferent co-payment interventions succeeding each other seem tohave curbed the upward trend in drug utilisation. Each interven-tion was succeeded by a more restrictive policy, and in that waythere was limited learning effect and the use and expenditures werekept down. The last policy, which introduced an even more re-strictive reimbursement for drugs, as well as a co-payment on out-patient services, and which removed coverage for over-the-counterdrugs, was the most intensive and had the largest effect on theoverall use of drugs. As expected, there was also a reduction in theuse of over-the-counter drugs. For all three groups, the reductionin use was larger than the reduction in expenditures. The investi-gators explained that this may have been due to patients obtaininglarger quantities of drugs per prescription or more expensive med-ications, since the fixed co-payment did not encourage people tochoose less expensive substitutions. Nevertheless, the investigatorsconcluded that there were overall reductions in drug use, becausethe lower utilisation more than offset the increases in expendituresper prescription (Harris 1990).
Ceilings
Restricting the maximum contribution reduced use of inhaled cor-ticosteroids by children in Manitoba, Canada (Kozyrskyj 2001).The reduction was greater for children with mild to moderateasthma than for children with severe asthma. No data were re-ported on the effects on expenditures, health or healthcare utilisa-tion. The quality of the evidence was low.
20Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Ceiling policies, when combined with full payment, as in Man-itoba, would be expected to be more intensive for low-incomeusers. Unfortunately, the study had serious limitations since thecontrol groups were only matched based on severity of asthma andnot by income. Also, the control groups had a different policy inbaseline than the experimental groups. However, the investigatorsreported that both income groups reduced their drug use com-pared to the control group which had full drug coverage. In thegroup with mild to moderate asthma, the reduction was greater forthe high-income group at one year post. In the group with severeasthma, there was no difference between income groups one yearpost. One explanation for the difference in utilisation between in-come groups was that the income-based policy was more intensivefor the high-income group (compared to the less restrictive policyin the pre period).Introducing a ceiling policy may potentially cause a hoarding ef-fect, particularly for patients with chronic conditions, who arelikely to exceed the ceiling. Although the study reported odds ra-tios, the paper also provided a graph with repeated measures datawhere peaks in drug use were observed towards the end of eachyear. Whether these peaks in utilisation were due to stockpiling orother factors was not addressed by the investigators.
Fixed co-payments with a ceiling
Fixed co-payment with ceiling reduced drug use. The income-based policies introduced in Australia (McManus 1996) had sub-stantial effects on drug use, whereas the effects of the policy in-troduced in Canada were modest (Poirier 1998). In both stud-ies, there were no or minor differences in effect between incomegroups. The small differences between income groups in Australiamay point to dose responsiveness to the income-based interven-tions (i.e. the effect of the intensity of intervention relative to in-come on drug use). Reductions were found in both “essential” andother drugs. No effects on drug expenditures, health or healthcareutilisation were reported. Few evaluations were included and theoverall quality of the evidence was low.In addition to the expected reductions in drug use and plan drugexpenditures, a ceiling policy combined with a fixed co-paymentpolicy is expected to provide a safety net for patients, and thus isa less intensive intervention than a fixed co-payment policy alone.However, it was not possible to compare these two policy groups,because of the heterogeneity of studies and difficulties in assessingthe intensity of different policies.There were several factors that may have modified the effect ofthese policies. Firstly, in the policy introduced in Australia, therepatriation group was given an allowance in addition to othersocial security entitlements (McManus 1996). This allowance wasindependent of prescription drug use and could also be used forother purposes. It is interesting to note that the policy had aneffect on this group, despite the fact that they in theory werefully compensated. A limitation of this study was that the study
population also included elderly patients who went from full drugcoverage to AUD 2.50 per prescription and who became eligiblefor a maximum co-payment level. Unfortunately, the authors didnot differentiate between the elderly and the rest of the communitywhen reporting the outcomes. One would expect a larger reductionin the elderly.The CAD 2 fixed co-payment with a ceiling in Quebec, Canadahad only a modest effect on drug utilisation in both income groupsand drug categories (Poirier 1998). The primary explanation maybe that drug use was more or less unchanged because the increasedpatient co-payments did not exceed the patients’ resources andneed for their medication, and that the insurer achieved savingsthrough a partial shift of costs to patients. However, the inves-tigators state that pharmacists expressed concerns regarding theintroduction of the co-payment policy, and may therefore havebeen willing to share some of the burden. Unfortunately no datawere reported on this. Finally, the study had limitations becauseno data were reported on the selection criteria or comparability ofthe control groups. Also, low- and high-income groups were de-fined based on postal codes, and the reliability of this is uncertain.None of the studies reported on potential stockpiling effects, whichmay be expected after patients have reached the ceiling. However,the time series graph provided in McManus et al (McManus 1996)shows strong patterns of seasonality towards the end of each yearpost policy, particularly for the repatriation group and in the useof “essential drugs”. Patterns of seasonality (potential hoarding)could also be seen in the time series for the community cohort.
Coinsurance with a ceiling
Coinsurance with a ceiling reduced the overall drug use andplan drug expenditures. Comparing the four different co-paymentplans introduced in the USA (RAND health insurance experi-ment), there was a dose response to the intensity of interventionswhere the greatest effect was found in the 50% and 95% coinsur-ance groups with income-based ceilings. The income-based pol-icy introduced in Canada reported on use of “essential” and otherdrugs, and found reductions in both drug categories, although thereduction appeared to be higher for other drugs. In the same studyit was found that low-income patients had the greatest reductionin drug use, despite the fact that they were subject to a less inten-sive co-payment. No data were reported on the effects on healthor use of healthcare services. Few evaluations were included andthe quality of evidence was moderate to low.Different effects were found for “essential” drugs. Blais et al (Blais2002) noted that the reduction in drug use was less prominentwith drugs that may be accompanied with more quickly perceiveddiscomfort by the patients.No data on health were provided, but Blais et al (Blais 2002) reportthat the reduction in use of “essential” drugs was associated witha higher rate of serious adverse events and emergency departmentvisits.
21Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Coinsurance policies, like fixed co-payments, are expected to re-duce drug use and induce a partial shift of cost from insurer topatient. This was supported by the evidence of the included stud-ies. Unlike fixed co-payments, coinsurance may also urge patientsto choose a cheaper drug, since the co-payment is based on a per-centage of the drug’s price. This was not evaluated in any of theincluded studies.Some factors may have modified the effects of the policies. Blaiset al (Blais 2002) reported that the large immediate effect of thepolicy may have been influenced by an observed hoarding effectthat could be seen immediately before the policy was implemented.The RAND health insurance experiment may have been a morerestrictive policy, since the coinsurance and ceiling also includedco-payment on services and not only drugs. Thus the patient notonly had to pay a certain amount per prescription but also had topay to see the physician to get a prescription. It was hypothesisedby the authors that people assigned to less generous insurance forprescription drugs would substitute with over-the-counter drugs.This was not the case. In fact, the study reported that people withmore complete prescription coverage used more of both groups ofdrugs than people with less coverage. The investigators interpretedthis as indicating that better financial access to care appeared topromote rather than substitute over-the-counter drug use. Therewere also observed differences in use between sites, suggesting thatsubstitution with self-care through over-the-counter drugs occurswhen access to physicians and formal care in rural areas is poor(Newhouse 1993).When combing coinsurance (or fixed co-payment) with a ceiling,the intensity of the policy works on two levels. The size of thecoinsurance and the ceiling will together determine how fast thepatient will reach the maximum contribution level. For example ahigh coinsurance, like in the 95% RAND coinsurance plan, maymake it easier to reach the ceiling. Likewise, a low ceiling may alsomake it more probable to reach the maximum contribution level.The probability of reaching the ceiling will also be dependent onthe size of other medical expenditures. Those with high medicalexpenditures will be more likely to reach the ceiling, and the size ofthe coinsurance may not have been so important for this particulargroup. In the RAND health insurance experiment, this was thecase. In other words, if the ceiling is easily reachable, the size ofthe coinsurance may not matter so much.None of the studies reported on potential hoarding effects afterpatients reached the ceiling. Although the Canadian interventionswere evaluated using RM design, it is not possible to detect possiblehoarding effects from the time series data since the policies werereplaced just four months later by other policies.
Fixed co-payments and coinsurance with aceiling
Introducing a fixed co-payment with coinsurance and a ceilinghad little or no effect on drug use in the included study and with
the exception of women’s use of antidepressants, there were nosustained effects on drug use (Ong 2003). Effects on health orhealthcare services were not reported. The quality of the evidencewas low. No data on expenditures were reported, but the inves-tigators stated that the policy resulted in a statistically significantincrease in pharmaceutical expenditures by Swedish consumers(Ong 2003).
Tier co-payments
Tier co-payment structures reduced drug use across all tiers, in-cluding drugs used for treating chronic illnesses. Based on the evi-dence from the included studies, three-tier co-payment structuressaved expenditures for the insurer by increasing the consumers’financial responsibility for prescription drugs and inducing a shiftof cost from the insurer to patients.Moving from a one-tier to a three-tier plan (with one-tier co-pay-ment increased by USD 1) had a greater effect on drug use andshift of cost than moving from a two-tier to a three-tier systemin one study (Huskamp 2005). The change in co-payment struc-ture induced a greater reduction in use of non-preferred branddrugs (third tier), than for preferred brand drugs (second tier).The effects on health are unclear. One study reported on use of ERand physician visits and found no statistically significant increases.However, large reductions in important drugs used for chronic ill-nesses, available in all tiers, were observed in the plan going froma one-tier to a three-tier system. This may have had unintendedeffects on patients’ health. Few evaluations were included and thequality of the evidence was low.Tiered co-payments are intended to prompt patients to choosemore cost-effective drugs or to cover the extra expenses themselves.However, tiered co-payments may also reduce overall drug useif patients are not willing to substitute for other drugs or if thechanges in the tier structure also include increased co-paymentsfor generics, as in the two interventions included here. AlthoughHuskamp et al (Huskamp 2005) did not include outcomes on thisthat met our inclusion criteria for study design (rate of switchingin baseline not reported), they report that among users of ACEinhibitors going from a one-tier to a three-tier system with a USD1 increase in co-payments for tier-one drugs, 42% continued touse the tier-three drug, 42% switched to cheaper drugs, and 16%discontinued the drug altogether. For statins, 31% of 83 patientscontinued using the tier-three drug, 53% switched to a lower tierand 23% discontinued the drug. The discontinuation rate wassomewhat lower in the plan moving from a two-tier to a three-tiersystem.Motheral et al (Motheral 2001) reported the effects on all threetiers and observed the largest reduction in the use of tier-threedrugs. A small reduction was also found in tier-two drugs, whileonly a minor and not statistically significant reduction was foundin generics. Although there might have been some substitutionfrom tier-three to tier-two and tier-one, the co-payment increases
22Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
for tier-one and tier-two may have counteracted the potentialswitching effects of the change in tier structure. The investigatorsalso note that patients may have continued their drug use, butpaid for their drugs outside of the formulary.Data on the effects on health and healthcare utilisation were onlyreported in the study that had a modest reduction in drug useacross tiers. The authors found small, statistically non-significantchanges in ER visits, hospital admissions and physician visits.However, the substantial reductions in the use of drugs consid-ered important and needed for the treatment of chronic illnessesby patients on an ongoing basis (ACE inhibitors and statins) inthe plan moving from a one-tier to a three-tier system, suggestspotentially harmful effects.Several factors may have modified the effects of the policies. Firstof all, two of the policies also had increases in co-payments forgenerics, and one of the plans moving from a two-tier to three-tieralso increased the co-payment for tier-two drugs. This may havecontributed to a more intensive policy and led to the reduction indrug use across tiers.A limitation in the evaluation by Huskamp et al (Huskamp 2005)was that the effects on the second and third tiers were reportedaggregated. Moreover, in the evaluation where the interventiongroup went from a one-tier to a three-tier system with a USD 1fixed co-payment increase for tier-one drugs, the comparison wasa two-tier system. The intervention and comparison groups haddifferent policies at baseline. This may have underestimated theeffects of the three-tier system compared to a one-tier system.Plan drug expenditures were reduced for most drug groups in allthree comparisons (Huskamp 2005; Motheral 2001). Rebates werenot included, which may have potentially further enhanced thesavings (by volume-based discounts for use of particular drugs).Only Motheral et al (Motheral 2001) reported data on long-termeffects; however, these had serious limitations since the entire pop-ulation was not included in the long-term follow up. Of the origi-nal population in the commercially insured preferred-provider or-ganisation, 8681 subjects were excluded from the original sampleof 13,279 in the second year because their employer made a changein the co-payment or pharmacy benefit plan. It is not known whysome employers chose to discontinue the three-tier plan.Finally, all three comparisons reported in these studies were basedin privately insured “healthy” populations. The effect of tier co-payment systems on low-income groups or other vulnerable pop-ulations is not known.
A U T H O R S ’ C O N C L U S I O N S
Implications for practice
Introducing direct co-payments reduced drug use across studies.Patients responded by discontinuing drugs or by paying an in-creased proportion of the costs themselves. Reductions were found
for life-sustaining drugs or drugs that are important in treatingchronic conditions as well as other drugs, suggesting that patientsmay not have been able to prioritise their drug use when facedwith a reimbursement restriction.
While the shift of cost from the insurer to patients in many casesled to savings for the insurer, the discontinuation of drugs mayhave had unintended effects. Few studies reported on the effectson health and healthcare utilisation. Possible adverse effects onhealth through increased healthcare utilisation were found whena cap was introduced in a vulnerable population (low-income pa-tients with chronic conditions). Other direct payment interven-tions may also have adversely affected patients through discon-tinuation of life-sustaining drugs or drugs that are important intreating chronic conditions. The deterioration of health in thesevulnerable populations may in the end result in increased use ofhealthcare services and overall plan expenditures.
Direct payments are less likely to cause harm if only non-essentialdrugs are included or exemptions are built in to ensure that pa-tients receive needed medical care. If direct payments for non-es-sential drugs are introduced to reduce overuse or misuse of drugs,it is questionable whether direct payments are the right way to ad-dress this, thereby continuing to partially cover inappropriate druguse. Other interventions, such as education or prior authorisation,might be better suited to address inappropriate use of drugs. Ifdirect payments are used by insurers to keep taxes or premiumsdown, the insurer should be clear about the rationale for the policyand consider risks of unintended effects on health and healthcareutilisation.
Most of the included evaluations were observational studies, someof which had wide confidence intervals. Few evaluations were in-cluded in each policy group and the quality of the evidence wasfound to be generally low to moderate.
Implications for research
Rigorous evaluation should be considered when implementing orintensifying direct payments for drugs. Potential effects on health,healthcare utilisation and administration costs should be mea-sured, preferably using RCT, RM or ITS designs. There is a needfor improving and standardising how the intensity of interven-tions is reported in such evaluations, particularly in relationship tothe disposable income of patients or some other metric that allowscomparisons across settings and studies.
A C K N O W L E D G E M E N T S
We gratefully acknowledge:
- Marit Johansen for conducting the literature searches
- Sue Hill, Malcolm Maclure, Carolyn J. Green and Kristin Kamilla
23Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Linnestad for sifting references and abstracts from the broad litera-ture search for pharmaceutical policies and commenting on draftsof the data collection form.
- Curt D. Furberg, Mark Gibson, Roberto Grilli, David A. Henry,Bob Nakagawa, Dennis Ross-Degnan, Gail Shearer, Stephen B.Soumerai, Luke Vale, Lisa Bero, Kirby Lee, Merrick Zwarensteinand Alain Mayhew provided helpful comments on drafts of theprotocol or the review, or both.
- Kjetil Olsen, Morten Bjørklund and Matthew Oxman, and theLibrary of the Norwegian Directorate for Health and Social Affairswho helped retrieve, copy and register studies.
R E F E R E N C E S
References to studies included in this review
Blais 2002 {published data only}
Blais L, Boucher J, Couture J, Rahme E, Lelorier J. Impactof cost-sharing drug insurance plan on drug utilizationamong older people. JAGS 2001;49:410–4.∗ Blais L, Couture J, Rahme E, LeLorier J. Impact of a costsharing drug insurance plan on drug utilization amongindividuals receiving social assistance. Health Policy 2002;64:163–72.
Brian 1974 {published data only}∗ Brian EW, Gibbens SF. California’s Medi-Cal CopaymentExperiment. State of California, Department of Health. J.B.Lippincott Company, US, 1974.
Cromwell 1999 {published data only}∗ Cromwell DM, Bass EB, Steinberg EP, Yasui Y, Racich WJ,Hendrich TR, et al.Can restrictions on reimbursement foranti-ulcer drugs decrease Medicaid pharmacy costs withoutincreasing hospitalizations?. Health Services Research 1999;33: 6:1593–610.
Donnelly 2000 {published data only}
Donnelly N, McManus P, Dudley J, Hall W. Impact ofincreasing the re-supply interval on seasonality of subsidisedprescription use in Australia. Australian and New Zealand
Journal of Public Health 2000;24(6):603–6.
Fryatt 1994 {published data only}
Fryatt RJ, Crowley SP, Gurung YB. Community financingof drug supplies in rural Nepal: Evaluating a “fee peritem” drug scheme. Health Policy and Planning 1994;9(2):193–203.
Harris 1990 {published data only}∗ Harris BL, Stergachis A, Ried LD. The effect of drug co-payments on utilization and cost of pharmaceuticals in ahealth maintenance organization. Medical Care 1990;28(10):907–17.
Huskamp 2005 {published data only}
Huskamp HA, Deverka PA, Epstein AM, Epstein RS,McGuigan KA, Frank RG. The effect of incentive-based
formularies on prescription-drug utilization and spending.New England Journal of Medicine 2003;349(23):2224–32.∗ Huskamp HA, Deverka PA, Epstein AM, Epstein RS,McGuigan KA, Muriel AC, et al.Impact of 3-tier formularieson drug treatment of attention deficit/hyperactivitydisorders in children. Archives of General Psychiatry 2005;62:435–41.
Hux 1997 {published data only}∗ Hux JE, Naylor CD, Fielding DA. The Ontario DrugBenefit Program Copayment: its impact on access forOntario seniors and charges to the program. Institute forClinical Evaluation Sciences in Ontario. Toronto, Arpil 24,1997.
Kozyrskyj 2001 {published data only}
Kozyrskyj AL, Mustard CA, Cheand MS, Simons FER.Income-based drug benefit policy: impact on receipt ofinhaled corticosteroid prescriptions by Manitoba childrenwith asthma. CMAJ 2001;165(7):897–902.
Lingle 1987 {published data only}∗ Lingle EWJ. The impact of outpatient drug benefits onthe use and costs of health care services for the elderly.Inquiry 1987;24(3):203–11.
Martin 1996 {published data only}∗ Martin BC, McMillan JA. The impact of implementing amore restrictive prescription limit on Medicaid recipients.Effects on cost, therapy, and out-of-pocket expenditures.Medical Care 1996;34:686–701.
McManus 1996 {published data only}
McManus P, Donnelly N, Henry D, Hall W, Primrose J,Lindner J. Prescription drug utilization following patientco-payment changes in Australia. Pharmacoepidemiology &
Drug Safety 1996;5(6):385–92.
Motheral 1999 {published data only}∗ Motheral BR, Henderson R. The effect of a copay increaseon pharmaceutical utilization, expenditures, and treatmentcontinuation. American Journal of Managed Care 1999;5(11):1383–94.
24Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Fairman KA, Motheral BR, Hendersen RR. Retrospective,long-term follow-up study of the effect of a three-tierprescription drug copayment system on pharmaceutical andother medical utilization and costs. Clinical Therapeutics
2003;25(12):3147–61.∗ Motheral B, Fairman KA. Effect of a three-tier prescriptioncopay on pharmaceutical and other medical utilization.Medical Care 2001;39(12):1293–304.
Newhouse 1993 {published data only}
Foxman B, Valdez RB, Lohr KN, Goldberg GA, NewhouseJP, Brook RH. The effect of cost sharing on the use ofantibiotics in ambulatory care: Results from a population-based randomized controlled trial. Journal of Chronic
Disease 1987;40(5):429–37.Leibowitz A. Substitution between prescribed and over-the-counter medications. Medical Care 1989;27(1):85–94.Leibowitz A, Manning WG, Newhouse JP. The demand ofprescription drugs as a function of cost-sharing. Soc. Sci.Med. 1985;21(10):1063–9.∗ Newhouse JP. Free for all? Lessons from the Rand HealthInsurance Experiment. US: Harvard University Press, 1993.Newhouse JP, Manning WG, Morris CN, Orr LL, DuanN, Keeler EB, et al.Some interim results from a controlledtrial of cost sharing in health insurance. The New EnglandJournal of Medicine 1981;305(25):1501–7.
Ong 2003 {published data only}∗ Ong M, Catalano R, Hartig T. A time-series analysis ofthe effect of increased copayments on the prescription ofantidepressants, anxiolytics, and sedatives in Sweden from1990 to 1999. Clinical Therapeutics 2003;25(4):1264–75.
Poirier 1998 {published data only}∗ Poirier S, LeLorier J, Page V, Lacour A. The effect of a $2co-payment on prescription refill rates of Quebec elderlyand its relationship to socio-economic status. Canadian
Pharmacy Journal 1998;131(1):30–4.
Reeder 1985 {published data only}
Nelson AA, Reeder CE, Dickson WM. The effect of aMedicaid drug copayment program on the utilization andcost of prescription services. Medical Care 1984;22(8):724–36.∗ Reeder CE, Nelson AA. The differential impact ofcopayment on drug use in a Medicaid population. Inquiry1985;22:396–403.
Sawyer 1982 {published data only}
Sawyer DO. Medicaid spending and state cost control:Assessing policy impact on drug expenditure. Health Policy
Quarterly 1982;2(2):98–123.
Soumerai 1994 {published data only}
Soumerai S, Ross-Degnan D. Unintended Outcomesof Health Care Policy and Research. Boston: HarvardUniversity, 1992:1–24.Soumerai SB, Avorn J, Ross-Degnan D, Gortmaker S.Payment restrictions for prescription drugs under medicaid.
Effects on therapy, cost, and equity. New England Journal of
Medicine 1987;317(9):550–6.Soumerai SB, McLaughlin TJ, Ross-Degnan D, Casteris CS,Bollini P. Effects of limiting Medicaid drug-reimbursementbenefits on the use of psychotropic agents and acute mentalhealth services by patients with schizophrenia. New EnglandJournal of Medicine 1994;331(10):650–5.∗ Soumerai SB, Ross-Degnan D, Avorn J, McLaughlin T,Choodnovskiy I. Effects of Medicaid drug-payment limitson admission to hospitals and nursing homes.[comment].New England Journal of Medicine 1991;325(15):1072–7.
Tamblyn 2001 {published data only}∗ Tamblyn R, Laprise R, Hanley JA, Abrahamowicz M,Scott S, Mayo N, et al.Adverse events associated withprescription drug cost-sharing among poor and elderlypersons [comment]. JAMA 2001;285(4):421–9.
References to studies excluded from this review
Andrade 2002 {published data only}
Andrade SE, Gurwitz JH. Drug benefit plans undermanaged care: To what extent do older subscribers selectingless drug coverage put themselves at increased financial risk?. Journal of the American Geriatrics Society 2002;50(1):178–81.
Biritwum 2001 {published data only}∗ Biritwum RB. Impact of health care financing on themanagement of malaria in Ghana. East African Medical
Journal 12 December 2001;78(12):636–40.
Chalker 1995 {published data only}
Chalker J. Effect of a drug supply and cost sharing systemon prescribing and utilization: a controlled trial from Nepal.Health Policy & Planning 1995;10(4):423–30.
Danzon 1997 {published data only}
Danzon PM, Liu H. Reference pricing and physiciandrug budgets: The German experience in controllingpharmaceutical expenditures. Wharton Business SchoolWorking Paper, University of Pennsylvania 1997.
Dor 2004 {published data only}
Dor A, Encinosa W. Does cost sharing affect compliance?The case of prescription drugs. NBER Working Paper No.10738; National Bureau of Economic Research, 2004.
Fortress 2001 {published data only}∗ Fortress EE, Soumerai SB, McLaughlin TJ, Ross-DegnanD. Utilization of essential medications by vulnerable olderpeople after a drug benefit cap: importance of mentaldisorders, chronic pain, and practice setting. Journal of the
American Geriatrics Society 2001;49(6):793–7.
Grootendorst 1997 {published data only}
Grootendorst PV. Health care policy evaluation usinglongitudinal insurance claims data: an application of thepanel Tobit estimator. Health Economics 1997;6:365–82.
Hall 1966 {published data only}
Hall CP. Deductibles in health insurance: an evaluation.Journal of Risk Insurance 1966;23:253–63.
25Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Hu S, Chen W, Cheng X, Chen K, Zhou H, Wang L.Pharmaceutical cost-containment policy: experiences inShanghai, China. Health Policy Plan 2001;16(2):4–9.
Johnson 1997 {published data only}
Johnson RE, Goodman MJ, Hornbrook MC, EldredgeMB. The effect of increased prescription drug cost-sharingon medical care utilization and expenses of elderly healthmaintenance organization members. Medical Care 1997;35(11):1119–31.
Lurk 2004 {published data only}
Lurk JT, DeJong DJ, Woods TM, Knell ME, Carroll CA.Effects of changes in patient cost sharing and drug samplepolicies on prescription drug costs and utilization in asafety-net-provider setting. American Journal of Health-System Pharmacy 2004;61(3):267–72.
Nair 2005 {published data only}
Nair KV, Valuck RJ, Allen RR, Lewis SJ. Impact of increasedcopayments on the discontinuation/switching rates ofnonformulary medications. Journal of Pharmacy Technology2005;21:137–43.
O’Brien 1989 {published data only}
O’Brien B. The effect of patient charges on the utilisation ofprescription medicines. Journal of Health Economics 1989;8(1):109–32.
Okunade 1998 {published data only}
Okunade A. Financing structure dynamics and otherdeterminants of Medicaid pharmaceutial expenditures,1993-1996. J Health Care Finance 1998;25(1):59–72.
Pilote 2002 {published data only}
Pilote L, Beck C, Richard H, Eisenberg MJ. The effects ofcost-sharing on essential drug prescriptions, utilization ofmedical care and outcomes after acute myocardial infarctionin elderly patients. CMAJ 2002;167(3):246–52.
Rector 2003 {published data only}
Rector TS, Finch MD, Danzon PM, Pauly MV, MandaBS. Effect of tiered prescription copayments on the use ofpreferred brand medications. Medical Care 2003;41(3):398–406.
Rudholm 2005 {published data only}
Rudholm N. Pharmaceutical insurance and the demandfor prescription pharmaceuticals in Västerbotten, Sweden.Scandinavian Journal of Public Health 2005;33:50–6.
Smith 1989 {published data only}
Smith MC, Simmons SA. A study of the effects of formularylimitations in Medicaid drug programs. The effectiveness of
medicines in containing health care costs: Impact of innovation,regulation and quality. The project HOPE workshop, TheNational Pharmaceutical council, 1989:117–41.
Starmans 1994 {published data only}∗ Starmans B, Janssen R, Schepers M, Verkooijen M. Theeffect of a patient charge and a prescription regulationon the use of antihypertension drugs in Limburg, TheNetherlands. Health Policy 1994;26(3):191–206.
Van Doorslaer 1984 {published data only}
Van Doorslaer E. The effects of cost sharing on the demandfor prescription drugs in Belgium. Acta Hospitalia 1984;3:69–81.
Weeks 1973 {published data only}
Weeks HA. Changes in prescription drug utilization afterthe introduction of a prepaid drug insurance program.Journal of the American Pharmaceutical Association 1973;13(4):205–9.
Winkelmann 2004 {published data only}
Winkelmann R. Co-payments for prescription drugs andthe demand for doctor visits - Evidence from a naturalexperiment. Health Economics 2004;13:1081–9.
References to studies awaiting assessment
Almarsdottir 2000 {published data only}∗ Almarsdottir AB, Morgall JM, Grimsson A. Costcontainment of pharmaceutical use in Iceland: the impactof liberalization and user charges. Journal of Health Services
Research and Policy April 2000;5(2):109–13.
Andersson 2006 {published data only}
Andersson K, Petzold MG, Sonesson C, Lonnroth K,Carlsten A. Do policy changes in the pharmaceuticalreimbursement schedule affect drug expenditures?Interrupted time series analysis of cost, volume and cost pervolume trends in Sweden 1986-2002. Health Policy 2006;79(2-3):231–43.
Chang 2005 {published data only}
Chang KC. Implementation of 3-tier plan design andemloyee education slows drug spend. Drug Benefit Trends
2005;17:162–71.
Dormuth 2006 {published data only}
Dormuth CR, Glynn RJ, Neumann P, Maclure M,Brookhart AM, Schneeweiss S. Impact of two sequentialdrug cost-sharing policies on the use of inhaled medicationsin older patients with chronic obstructive pulmonary diseaseor asthma. Clin Therapy 2006;28(6):964–78.
Gibson 2005 {published data only}
Gibson TB, McLaughlin CG, Smith DG. A copaymentincrease for prescription drugs: the long-term and short-term effects on use and expenditures. Inquiry 2005;42(3):293–310.
Holloway 2001a {published data only}
Holloway KA, Gautam BR, Reeves BC. The effects ofdifferent kinds of user fee on prescribing costs in ruralNepal. Health Policy Plan 2001 Dec;16(4):421–7.
Holloway 2001b {published data only}
Holloway KA, Gautam BR, Reeves BC. The effects ofdifferent kinds of user fees on prescribing quality in ruralNepal. J Clin Epidemiol 2001 Oct;54(10):1065–71.
Kephart 2007 {published data only}
Kephart G, Skedgel C, Sketris I, Grootendorst P, Hoar J.Effect of copayments on drug use in the presence of annualpayment limits. American Journal of Managed Care 2007;6II:328–334.
26Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Landsman PB, Yu W, Liu X, Teutsch SM, Berger ML.Impact of 3-tier pharmacy benefit design and increasedconsumer cost-sharing on drug utilization. The AmericanJournal of Managed Care 2005;11(10):621–8.
Lee 2006 {published data only}
Lee YC, Yang MC, Huang YT, Liu CH, Chen SB.Impacts of cost containment strategies on pharmaceuticalexpenditures of the National Health Insurance in Taiwan,1996-2003. Pharmacoeconomics 2006;24(9):891–902.
Liu 2004 {published data only}
Liu SZ, Romeis JC. Changes in drug utilization followingthe outpatient prescription drug cost-sharing program -evidence from Taiwan’s elderly. Health Policy 2004;68:277–87.
Additional references
Aaserud 2003Aaserud M, Dahlgren AT, Sturm H, Kösters JP, Hill S,Furberg CD, et al.Policies: effects on rational drug use, anoverview of 13 reviews. (Protocol). Cochrane Database ofSystematic Reviews 2003, Issue 2. [Art. No.: CD004397.DOI: 10.1002/14651858.CD004397.pub2]
Aaserud 2006Aaserud M, Dahlgren AT, Kösters JP, Oxman AD, RamsayC, Sturm H. Pharmaceutical policies: effects of referencepricing, other pricing, and purchasing policies. CochraneDatabase of Systematic Reviews 2006, Issue 2. [Art. No.:CD005979. DOI: 10.1002/14651858.CD005979]
Adams 2001Adams AS, Soumerai S, Ross-Degnan D. The case for amedicare drug coverage benefit: a critical review of theempirical evidence. Annual Review of Public Health 2001;22:49–61.
Davey 2005Davey P, Brown E, Fenelon L, Finch R, Gould I, HolmesA, et al.Interventions to improve antibiotic prescribingpractices for hospital inpatients. Cochrane Database ofSystematic Reviews 2005, Issue 4. [Art. No.: CD003543.DOI: 10.1002/14651858.CD003543.pub2]
Dewa 2003Dewa CS, Hoch JS. Fixed and flexible formularies as cost-control mechanisms. Expert Review of Pharmacoeconomics
and Outcomes Research 2003;3(3):303–15.
Domino 2003Domino ME, Salkever DS. Price elasticity andpharmaceutical selection: the influence of managed care.Health Economics 2003;12(7):565–86.
EPOC 2002Cochrane Effective Practice and Organisation of Care ReviewGroup. THE DATA COLLECTION CHECKLIST. http://www.epoc.uottawa.ca/checklist2002.doc. Accessed Feb 22,2006.
Ess 2003Ess SM, Schneeweiss S, Szucs TD. Europeanhealthcare policies for controlling drug expenditure.Pharmacoeconomics 2003;21(2):89–103.
Freemantle 1996Freemantle N, Bloor K. Lessons from internationalexperience in controlling pharmaceutical expenditure. I:Influencing patients. BMJ 1996;312(7044):1469–71.
Gibson 2003Gibson TB, Ozminkowski RJ, Goetzel RZ. The effects ofprescription drug cost sharing: a review of the evidence.American Journal of Managed Care 2003;11(11):730–40.
Ginsburg 1973Ginsburg PB, Manheim LM. Insurance, co-payment, andhealth services utilization: A critical review. Journal ofEconomics and Business 1973;25:142–53.
Gleason 2004Gleason PP, Solberg CL, Brelje TV, Hartwig SC, Wadd W.Assessment of a retrospective drug utilization review alertletter program. Journal of Clinical Outcomes Management
2004;11:150–4.
GRADE 2004The GRADE working group. Grading quality of evidenceand strength of recommendations. BMJ 2004;328(7454):1490–4.
Griffin 1996Griffin JP. An historical survey of UK government measuresto control the NHS medicines expenditure from 1948 to1996. Pharmacoeconomics 1996;10(3):210–24.
Grilli 2002Grilli R, Ramsay CR, Minozzi S. Mass media interventions:effects on health services utilisation. Cochrane Database of
Gross 1994Gross DJ, Ratner J, Perez J, Glavin SL. Internationalpharmaceutical spending controls: France, Germany,Sweden, and the United Kingdom. Health Care Financing
Review 1994;15(3):127–40.
Haaijer-Ruskamp 2002Haaijer-Ruskamp FM. Experience with patient charges.International Journal of Risk & Safety in Medicine 2002;15:93–6.
Harten 2004Harten C, Ballantyne P. The impact of cost-sharing withinCanadian provincial drug benefit programs: a review.Journal of Pharmaceutical Finance, Economics & Policy 2004;13:35–53.
Huttin 1994Huttin C. The use of prescription charges. [Review]. HealthPolicy 1994;27:53–7.
Jang 1988Jang R. Medicaid formularies: a critical review ofthe literature. Journal of Pharmaceutical Marketing &
Management 1988;2(3):39–61.
27Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Johnston 1991Johnston M. The price elasticity of demand forpharmaceuticals. Economics and Health: Proceedings ofthe Twelfth Australian conference. 1990.
Levy 1992Levy RA. Prescription cost sharing: economic andhealth impacts, and implications for health policy.Pharmacoeconomics 1992;2(3):219–37.
Lexchin 2002Lexchin J, Grootendorst P. The effects of prescription druguser fees on health services use and health status: a reviewof the evidence. Working Paper, Department of Clinicaland Epidemiology and Biostatistics. McMaster University,Ontario, Canada. Ontario, Canada, 2002.
Lyles 1999Lyles A, Palumbo FB. The effect of managed careon prescription drug costs and benefits. [Review].Pharmacoeconomics 1999;15(2):129–40.
Mossialos 2004Mossialos E, Walley T, Mrazek M. Regulatingpharmaceuticals in Europe: an overview. Regulatingpharmaceuticals in Europe: striving for efficiency, equity and
quality. Open University Press, 2004:1–37.
Nair 2004Nair KV, Valuck RJ. Impact of three-tier pharmacy benefitstructures on consumer attitudes, pharmacy, medicalutilization and costs: a critical review. Disease Management& Health Outcomes 2004;12:81–92.
Narine 1997Narine L, Senathirajah M. Pharmaceutical cost containmentpolicies: intended and unintended impacts. Department of
Health Administration, Faculty of Medicine. University ofToronto, 1997.
Noyce 2000Noyce PR, Huttin C, Atella V, Brenner G, Haaijer-RuskampFM, Hedvall M, et al.The cost of prescription medicines topatients. Health Policy 2000;52(2):129–45.
Ramsay 2003Ramsay CR, Matowe L, Grilli R, Grimshaw JM, ThomasRE. Interrupted time series design in health technology
assessment: Lessons from two systematic reviews of behaviorchange strategies. Interrupted time series design in healthtechnology assessment: Lessons from two systematic reviews of
behavior change strategies 2003;19(4):612–23.
Reeder 1993Reeder CE, Lingle EW, Schulz RM, Mauch R, NightengaleBS, Pedersen CA, et al.Economic impact of cost-containment strategies in third party programmes in the US(Part 1). Pharmacoeconomics 1993;4(2):92–103.
Reutzel 1993Reutzel T. The nature and consequences of policies intendedto contain costs in outpatient drug insurance programs.Clinical Therapeutics 1993;15(4):752–64.
Rice 1994Rice T, Morrison KR. Patient cost sharing for medicalservices: a review of the literature and implications forhealth care reform. Medical Care Review 1994;51(3):235–87.
Rice 2004Rice T, Matsuoka KY. The impact of cost-sharing onappropriate utilization and health status: a review of theliterature on seniors. Medical Care Research and Review2004;61:415–52.
Smith 1992Smith DG, Kirking DM. Impact of consumer fees on drugutilisation. Pharmacoeconomics 1992;2(4):335–42.
Soumerai 1987Soumerai SB, Avorn J, Ross-Degnan D, Gortmaker S.Payment restrictions for prescription drugs under medicaid.Effects on therapy, cost, and equity. New England Journal of
Medicine 1987;317(9):550–6.
Soumerai 1990Soumerai S, Ross-Degnan D. Experience of state drugbenefit programs. Health Affairs 1990;9(3):36–54.
Thomson 2004Thomson S, Mossialos E. Influencing demand for drugsthrough cost sharing. Regulating pharmaceuticals in Europe:striving for efficiency, equity and quality.. Open UniversityPress, 2004:227–44.
∗ Indicates the major publication for the study
28Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Outcomes Drug useHealthcare utilisationPlan drug expenditures
Notes Supported by National Research Service Award from the Agency for Health Care Policy and Research,and by educational grant from Hoechst Marion Roussel, Inc
Outcomes Drug useHealthcare utilisationHealthcare expenditures (subsidy needed)
Notes Source of funding not reported
Risk of bias
Item Authors’ judgement Description
Allocation concealment? Unclear D - Not used
Harris 1990
Methods CBASome limitations
Participants Setting: USA, Group health cooperative HMO
Interventions Fixed co-payment/fixed co-payment with cap
Outcomes Drug useTotal/plan drug expenditure
Notes Source of funding not reported
Risk of bias
Item Authors’ judgement Description
Allocation concealment? Unclear D - Not used
Huskamp 2005
Methods CBANo/some limitations
Participants Setting: USA, two large employer plans contracting a large health insurer
Interventions Tier co-payment
Outcomes Drug useTotal/patient/plan drug expenditures
Notes Supported by the Robert Wood Johnson Foundation’s Changes in Health Care Financing and OrganizationInitiative, Princeton, NJ /The National Institute of Mental Health, Bethesda, Md/Agency for HealthcareResearch and Quality, Rockville, Md./Alfred P. Sloan Foundation, New York, NY
31Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Participants Setting: USA, New Hampshire and New Jersey Medicaid/Medicare
Interventions Cap
Outcomes Drug useHealthcare utilisationPlan drug expenditures
Notes Supported by grants from the Agency for Healthcare Policy and Research, Department of Health andHuman Services/The John A. Hartford Foundation/The National Institute on Aging
Risk of bias
Item Authors’ judgement Description
Allocation concealment? Unclear D - Not used
Tamblyn 2001
Methods RMNo limitations
Participants Setting: Canada, Quebec,Quebec health insurance program (RAMQ)
Interventions Coinsurance with ceiling
37Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Notes Supported by the Ministry of Health in Quebec and the Règie de l’assurance maladie du Québec , Ministerede la Santé et des Services Sociaux, the Medical Research Council, the National Research DevelopmentProgram
Risk of bias
Item Authors’ judgement Description
Allocation concealment? Unclear D - Not used
Characteristics of excluded studies [ordered by study ID]
Study Reason for exclusion
Andrade 2002 Very serious limitations, untrustworthy results
Biritwum 2001 Interrupted time series, not enough data points
Chalker 1995 Cross sectional
Danzon 1997 Multiple policies measured simultanously, not possibly to measure the effect of co-payment
Dor 2004 Not interrupted time series
Fortress 2001 Incorrect analysis of timeseries, reanalysis of data not possible
Grootendorst 1997 Not interrupted time series
Hall 1966 Very serious limitations, untrustworthy results
Hu 2001 Not enough datapoints in the ITS analysis and quarterly data unavailable
Johnson 1997 Cross sectional
Lurk 2004 Single site/one-off intervention
Nair 2005 Time series with no baseline data
O’Brien 1989 Not interrupted time series
Okunade 1998 Not interrupted time series
38Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 1. Factors that could modify the effects of co-payment or cap policies
FACTOR CONDITION POTENTIAL EFFECTS(IF CONDITION NOT FULFILLED)
Size of cap/co-payment Reasonable direct patient payment/restrictionsfor drugs. The effect of the policy will be depen-dent on how strict the co-payment policy is (i.e. size of co-payment, time and volume includedin cap, period of time included in maximum co-payment polices etc)
If too strict:Drug use: DecreaseHealth: DecreaseHealthcare utilisation: IncreasePatient drug expenditures: IncreaseInsurer drug expenditures: DecreaseOther insurer expenditures: Increase
If too generous:Drug use: No changeHealth: No changeHealthcare utilisation: No changePatient drug expenditures: No change/increaseInsurer drug expenditures: No change/decreaseOther insurer expenditures: No change
Drug groups Suitable drug groups included in policy. Most co-payment policies usually aim to reduce the over-use of drugs and to control expenditures, target-ing particularly “less-essential” drugs. However,when co-payments are applied to all prescriptiondrugs in general, there is also a risk of discontin-uation of important drugs
Drug use: DecreaseHealth: DecreaseHealthcare utilisation: IncreasePatient drug expenditures: IncreaseInsurer drug expenditures: DecreaseOther insurer expenditures: Increase
Vulnerable populations Ensuring that all patients have access to, and canafford important life sustaining drugs. Low-in-come groups are sensitive towards co-payments,and increased drug expenses may cause discon-tinuation of important drugs. Not taking thispopulation into consideration, the policy may re-sult in higher healthcare utilisation, deteriorationof health, and higher overall healthcare expendi-tures.
Low-income chronics that are in need of mul-tiple drugs may be particular vulnerable to co-payment policies since they are more likely to ex-ceed any cap levels, or of using a high number of
Drug use: DecreaseHealth: DecreaseHealthcare utilisation: IncreasePatient drug expenditures: IncreaseInsurer drug expenditures: DecreaseOther insurer expenditures: Increase
40Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 1. Factors that could modify the effects of co-payment or cap policies (Continued)
drugs that may add up to large co-payments
Enforcement Adequate incentives for enforcer to comply withthe policy. Co-payments are in most cases en-forced by pharmacists or the physician. In somecases the pharmacist or physician have the powerto excempt patients from the co-payments, butwill then be liable for these expenses themselves.
If such an exemption is easy attainable for the pa-tient, little reduction in drug use can be expectedalthough the policy may still save third-party ex-penditures. Instead of a shift of cost from insurerto patient, there is a shift of cost from insurer toenforcer
Drug use: No changeHealth: No changeHealthcare utilisation: No changePatient drug expenditures: No change/increaseInsurer drug expenditures: No change/decreaseOther insurer expenditures: No change
Patient’s level of information Adequate follow-up and information provided topatients. Since many co-payment policies expectpatients to prioritise use of important and lifesustaining drugs over “less essential drugs”, muchresponsibility is put on the patient to make goodchoices about their own health and knowledgeof pharmacotherapy. However, without enoughinformation, patients may choose differently, forexample prioritising “less essential” drugs butwhich are associated with more rapidly experi-enced discomfort if discontinued
Drug use: Decrease (of important drugs)Health: DecreaseHealthcare utilisation: IncreasePatient drug expenditures: IncreaseInsurer drug expenditures: decreaseOther insurer expenditures: Increase
Enforcer’s level of information Adequate follow up and information provided toenforcer. How much information prescribers orpharmacists have about the policy when drugs areprescribed or dispensed, and how involved thepatient is in the decision making, are importantfactors.
Also, to what extent the enforcer is informedabout the price of drugs, drug substitution pos-sibilities or the patients’ ability to pay may influ-ence the impact of the policy.
Potential consequences may be that the use ofdrugs is unchanged, and that further economicstrain is put on the patient (instead of, for exam-ple, substituting for less expensive drugs)
Drug use: No changeHealth: No changeHealthcare utilisation: No changePatient drug expenditures: IncreaseInsurer drug expenditures: decreaseOther insurer expenditures: No change
Exemptions Reasonable mechanisms for patients that needexemptions for medical reasons. However, toogenerous exemptions may minimise the potentialeffects of the policy
If too strict:Drug use: DecreaseHealth: DecreaseHealthcare utilisation: Increase
41Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 1. Factors that could modify the effects of co-payment or cap policies (Continued)
Patient drug expenditures: IncreaseInsurer drug expenditures: DecreaseOther insurer expenditures: Increase
If too generous:Drug use: No changeHealth: No changeHealthcare utilisation: No changePatient drug expenditures: No change/ increaseInsurer drug expenditures: No change/ decreaseOther insurer expenditures: No change
If too generous:Drug use: No changeHealth: No changeHealthcare utilisation: No changePatient drug expenditures: IncreaseInsurer drug expenditures: DecreaseOther insurer expenditures: No change
Table 2. Drug categories included in studies
STUDY “ESSENTIAL” DRUGS “LESS ESSENTIAL/DISCRETIONARY/OTHER”DRUGS
Brian 1974 Critical:Drugs usually prescribed for serious conditions, nor-mally not cured but controlled by their use. The effectsof not taking these drugs are immediate, dramatic/orespecially traumatic
Needed, but not critical:Drugs usually prescribed for serious or almost seriousconditions, and where absence would indicate lack ofneeded care. However, the effects of not taking thesedrugs would not be immediate and dramatic althoughthe condition might deteriorate gradually
Broad range:Drugs prescribed for conditions of all levels of serious-ness which usually fit the definition of a needed drug,although in some instances they are of marginal need.In these latter instances, their absence does not indicatelack of needed care.
Preventive:Drugs whose use is preventive, primarily birth controldrugs
All other drugs:Drugs not classified into any of the above mentionedcategories
42Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
McManus 1996 Antiasmatic inhalantsAntiepilepticsAntithrombotic agentsAntiparkinson drugsAntihypertensivesAntiglaucoma preparations and mioticsBeta-blocking agentsCardiac glycosidesCorticosteroids for systemic use, plainDiureticsDrugs used in diabetes
AntacidsAntihistamines for systemic useAnti-inflammatory products, nonsteroidAntiobesity preparationsCough and cold preparationsHypnotics/sedativesOther analgesics and antipyreticsThroat preparationsVitamins
43Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Cough and cold remediesSkeletal muscle relaxantsNSAIDSAnalgesics
Motheral 2001 - -
Newhouse 1993 - -
Ong 2003 - AntidepressantsAnxiolyticsSedatives
Poirer 1998 Antihypertensives Benzodiazepines
Reeder 1985 Cardiovascular productsDiuretics
AnalgesicsSedative/hypnotics
Categories reported but not classified:AdrenergicsAntihistaminesAnti-infectivesCholinergicsGastrointestinalsPsychotherapeutics
Roemer 1975 - -
Sawyer 1982 - -
Soumerai 1987 InsulinPropranololThiazidesFurosemideMethyldopaLithiumDigoxinAnxiolytic and hypnotic agentsDepressants and lithiumAntipsychotics
Symptomatic relief drugs:AnalgesicsAntinflammatoryAspirinAcetaminophenPropoxyphene (in combination with aspirin or ac-etaminophen)Ibuprofen
Marginal or no efficacy drugs :Ergoloid mesylatesBarbiturate-anticholinergic combination agent Donna-talPropoxyphene without aspirin or acetaminophenAnticholinergic dicyclomine
44Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 4. Assessment of limitations in included studies - part 2 (Continued)
tween low-and high-in-come poli-cies in theoutcome10.24% loss tofollow up atthree yearspost11. Co-pay-ment on ser-vices imple-mented si-multane-ously12. Not re-ported13. Not allinstitutionswere able tosubmit theirclaims14. Differ-ences foundbetweencontrol andexperimen-tal group15. Notsame policyfor experi-men-tal and con-trol in base-line16.61% loss tofollow up17.28% loss tofollow up18.Only differ-ences are re-ported, notbaseline fig-
60Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 4. Assessment of limitations in included studies - part 2 (Continued)
large loss tofollow up attwo-yearpost-pol-icy, 8681 of13,279 wereex-cluded fromthe sam-ple becausetheir em-ployer madea change inthe co-payment orbenefit plan26. Not metfor statinsand ACEinhibitors:Differ-ence in usebetween ex-perimen-tal and con-trol groups
Table 5. Intervention description
STUDY ID/INTERVENTION PE-RIOD INSTUDY/STUDY POP-ULATION
INTERVENTION COMPARISON EXEMPTIONS/MODIFIERS
DRUGS INCLUDEDIN POLICY
POLICY:CAP
Cromwell 1999
2/1992- 12/1993
One anti-ulcer drug pre-scribed with only one re-fill.
Coverage for high doseprescription treat-ment for acute disorderswas limited to 60 days
No restrictions Not reported Anti-ulcers
62Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
20 days minimum re-supply cap for drugswith five or more repeats,pharmaceutical benefitsscheme items and eyedrops four days
Three day minimum re-supply cap for all drugs
Pharmacists could over-ride the re-supply rule
All prescription drugs
Martin 1996
11/1991- 5/ 1992
Five prescriptions reim-bursed per month
Other:Cough and cold drugswere taken off formulary(11/1991)
Benzodi-azepines on prior autho-risation (1/1992)
Six prescriptions reim-bursed per month
Not reported All prescription drugs
Soumerai 1994
9/1981- 7/1982
Three prescriptions re-imbursed per month
Other:Allowable quantity oftablets/capsules per pre-scription was tripled to amaximumthree-month supply
No restrictions Not reported All prescription drugs
POLICY:FIXEDCO-PAYMENTS
Brian 1974
1/1972- 6/1973(but measured at 3/1972post)
USD 0.50 fixed co-pay-ment per prescription forthe first two prescrip-tions (outpatients). Longterm inpatients must payUSD 0.50 co-paymentfor all prescriptions
Other:USD 1.00 co-paymentper visit for the two firstprovider visits
Benefits covering healthservices, but only twophysician visits/prescrip-tions/uses of lab-tests/x-rays per month
Prior authorisation onservices in both pre andpost policy periods
Patients eligible for bothMedi-cal and Medicarewere exempted from co-payments on Medi-calservices
All prescription drugs
Fryatt 1994
1/1991- 5/1991
USD 0.03 or USD 0.10 fixed co-payment perprescription
Fixed co-payment perprescription of USD 0.75 or USD 0.25 per pre-
Pregnant women, chil-dren under five and verypoor only charged the
All prescription drugs
63Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Full drug coverage Childrenand selected drug groupswere exempted from pol-icy (drugs for STDs andTB)
All prescription drugs
Tamblyn 2001Intervention 1
8/1996- 12/1996
See Blais 2002, interven-tion 1
See Blais 2002, interven-tion 1
See Blais 2002, interven-tion 1
See Blais 2002, interven-tion 1
Tamblyn 2001Intervention 2
8/1996- 12/1996
See Blais 2002, interven-tion 2
See Blais 2002, interven-tion 2
See Blais 2002, interven-tion 2
See Blais 2002, interven-tion 2
Newhouse 1993Intervention 1
Enrollmentbegan in Dayton Ohioin 1974 and 1975. Seat-tle, Washington; Fitch-burg and Leominster,Massachusetts; FranklinCounty, Massachusetts;Charleston, South Car-olina; and GeorgetownCounty, South Carolinaenrolled in 1977. Fami-lies participated for threeor five years
25% coinsurance ondrugs
Annual family incomebased ceiling of 5%,10% or 15%, or maxi-mum USD 1000
Other:25% coinsurance on in-patient and outpatientservices
Full drug and servicescoverage
All participantswere guaranteed not tobe “worse off ” than theprevious plan and wastherefore in some casescompensated by initia-tive payments (IP)
All prescription drugs
Newhouse 1993Intervention 2
Same as intervention 1
50% coinsurance ondrugs
Annual family incomebased ceiling of 5%,10% or 15%, or maxi-mum USD 1000
Other:50% coinsurance on in-patient and outpatientservices
Full drug and servicescoverage
All participantswere guaranteed not tobe “worse off ” than theprevious plan and wastherefore in some casescompensated by initia-tive payments (IP)
All prescription drugs
67Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
DDD Defined Daily Doses. A standardised concept established by the WHO. A DDD is the assumed average maintenancedose per day for a drug used for its main indication in adults
EPOC Effective Practice and Organisation of Care
ER Emergency room
ITS Interrupted time series
NSAID Non-Steroidal Anti-Inflammatory Drug
OECD Organisation for Economic Co-operation and Development
PPI Proton-Pump Inhibitor
PUD Peptic ulcer disease
RCT Randomised controlled trial
RM Repeated measures
RR Risk ratio (intervention vs control group)
RR (adj) Risk ratio (adjusted for pre intervention differences) = RR post intervention / RR pre intervention
SEK Swedish crowns
USD USA dollars
WHO World Health Organisation
70Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
pact on our confidencein the estimate of effectand is likely to changethe estimate.
Very low quality: We arevery uncertain about theestimate.
FOOTNOTES:1. Immediate or short-term effects. Reported aschange in level as thedifference between thefitted post value minuspredicted value (ITS)and adjusted relative risk(CBA)2. Only anti-ulcers included in policy(Cromwell 1999)3. Reportedas mean changes in levelfrom baseline (Soumerai1994).4. (Martin 1996)5. Anti-ulcers6. Patients using less pre-scriptions than includedin cap (prior to the inter-vention)7. Patients using moreprescriptions than in-cluded in cap (prior tothe intervention)8. Some uncertainty ofdirectness, only studiesfrom one setting in-cluded9. Symptomatic reliefdrugs10. Limited efficacydrugs11. Range by conditioncategory, com-plicated, uncomplicatedpeptic-ulcer disease, and
73Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
non-ulcer peptic condi-tion12. Partial hospi-talisations at communitymental health centers13. Severly disabledschizophrenics14. Elderly15. Nursing home ad-missions16. Visits to mentalhealth centers17. Effect at 12 monthswas -39.6 (-49.0%, -30.3%)18. Effect at 12 monthswas -32.0% (-40.7%,-23.3%)19. Effect at 12 monthsranged from 0.8% to 15.4% (all were statisticallynot significant)
Table 8. Summary of findings: Fixed co-payments
GRADE Evidence Pro-file
Question: Should fixedco-payments (versus fulldrug coverage) be usedfor rational drug use?
Patient or population:Low-income public in-surance program (Med-icaid) and privately in-sured
Settings: USA andCanada
Outcome(no of interventions)
Intervention Relative change (P value)[1]
Relative change (P value)[1]
Quality
74Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 8. Summary of findings: Fixed co-payments (Continued)
change our confidence inthe estimate of effect.
Moderate quality: Fur-ther research is likely tohave an important im-pact on our confidencein the estimate of effectand may change the esti-mate.
Low quality: Further re-search is very likely tohave an important im-pact on our confidencein the estimate of effectand is likely to changethe estimate.
Very low quality: We arevery uncertain about theestimate.
FOOTNOTES:1. Immediate or shortterm effects. Reported aschange in level as thedifference between thefitted post value minuspredicted value (ITS)and adjusted relative risk(CBA)2. 2 CAD for low-in-come, 100 CAD forinitial co-payment afterwhich patients paid 6.11CAD per Rx for high-in-come (Hux 1997)3. Short term effects notavailable, all results werereported as one year post(Harris 1990)4. 5 USD for genericsand 15 USD for branddrugs versus 4 USD forgenerics and 10 USD forbrand drugs (Motheral1999)
77Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 8. Summary of findings: Fixed co-payments (Continued)
5. 7 USD for genericsand 15 USD for branddrugs versus 5 USD forgenerics and 10 USD forbrand drugs (Motheral1999)6. (Reeder 1985)7. P-value not reported8. The absoluteeffect was -0.3 (P<0.05)monthly drug claims perpatient.9. Low-income10. Some uncertaintyof directness, only stud-ies from one setting in-cluded11. The interventionalso had $1 fixed co-payment per two firstprovider visits (Brian1974)12. Low-income elderly,disabled and familieswith dependent children13. Range of drug cat-egories defined as “crit-ical” and “needed” byBrian el al (1974)14. Reported as% change from baselineto post policy comparedto control15. Range of drug cate-gories defines as “broadrange”, “preventive” and“all other” by Brian el al(1974)16. Policy also includedrestrictions on reim-bursement for over thecounter drugs. The re-sult is reported as the ad-justed (for MAC policy)level effect from baseline(Sawyer 1982)17. Baseline data not re-ported. The absolute ef-fect was -2.0 (P<0.05)
78Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 9. Summary of findings: Fixed co-payments with ceiling (Continued)
pact on our confidencein the estimate of effectand is likely to changethe estimate.
Very low quality: We arevery uncertain about theestimate.
FOOTNOTES:1. Immediate or shortterm effects. Reported aschange in level as thedifference between thefitted post value minuspredicted value (ITS)and adjusted relative risk(CBA)2. Community: $15 perRx with ceiling versus$11 per Rx with ceil-ing post policy. Elderlyand social security: $2.5 per Rx with ceilingversus full drug cover-age. Level of ceiling wasnot reported (McManus1996)3. Only long term effectsreported (18 monthspost) (Poirer 1998).4. Level of ceiling wasnot reported (McManus1996)5. The absolute effectwas -816000 prescrip-tions dispensed (CI 95%;-1116133, -516373)6. Elderly with high-in-come7. The absolute effectwas -29500 prescrip-tions dispensed (CI 95%;-45812, -13287)8. Elderly with low-in-come9. Antihypertensives
81Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 10. Summary of findings: Coinsurance with ceiling (Continued)
95% coinsurance& annual family incomebased ceiling of 5%,10%or 15% [2]
-37.6% (P > 0.05)
Health(0)
- - - -
Healthcare utilisation(0)
- - - -
GRADE Work-ing Group grades of evi-dence
High quality: Further re-search is very unlikely tochange our confidence inthe estimate of effect.
Moderate quality: Fur-ther research is likely tohave an important im-pact on our confidencein the estimate of effectand may change the esti-mate.
Low quality: Further re-search is very likely tohave an important im-pact on our confidencein the estimate of effectand is likely to changethe estimate.
Very low quality: We arevery uncertain about theestimate.
FOOTNOTES:1. Immediate or shortterm effects. Reported asrelative risk (RCT) andchange in level as the dif-
84Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 10. Summary of findings: Coinsurance with ceiling (Continued)
ference between the fit-ted post value minus pre-dicted outcome (ITS)2. Only long term effectsreported (2< years post).Policy also included ser-vices (Newhouse 1993).3. Some uncertainty ofdirectness, only studiesfrom one setting in-cluded4. (Tamblyn 2001)5. Elderly with mixed in-come (but not very low)6. Welfare and low-in-come elderly
Table 11. Summary of findings: Tier co-payments
GRADE Evidence Pro-file
Question: Should tieredco-payments be used forrational drug use?
Patient or population:Privately insured
Settings: USA
Outcome(no of interventions)
Interventions Relative change (P value)[1]
Relative change (P value)[1]
Quality
Interrupted time seriesstudies
Controlled before-afterstudies
Overall drug use acrosstiers(3)
From 1- to 3-tier &1-tier co-payment in-creased by USD 1 [3,4]
-24.0% (P<0.001)5, -24.0% (P < 0.001) [6]
Very low [7]
From 2- to 3-tier [3] -5.0% (P = 0.26)5, -2.0% (P = 0.69) [6]
85Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 11. Summary of findings: Tier co-payments (Continued)
GRADE Work-ing Group grades of evi-dence
High quality: Further re-search is very unlikely tochange our confidence inthe estimate of effect.
Moderate quality: Fur-ther research is likely tohave an important im-pact on our confidencein the estimate of effectand may change the esti-mate.
Low quality: Further re-search is very likely tohave an important im-pact on our confidencein the estimate of effectand is likely to changethe estimate.
Very low quality: We arevery uncertain about theestimate.
FOOTNOTES:1. Immediate or shortterm effects, reported aschange in level as thedifference between thefitted post value minuspredicted outcome (ITS)and adjusted relative risk(CBA)2. All results are re-ported as one year post(Motheral 2001)3. All results are reportedas change in probabilitycompared to control atone year post (Huskamp
87Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
Table 11. Summary of findings: Tier co-payments (Continued)
2005)4. Control group had a2-tier system5. Ace- inhibitors (avail-able all tiers)6. Statins (available alltiers)7. Some uncertainty ofdirectness, only studiesfrom one setting in-cluded8. 2- tier (preferred)brand drugs9. 3- tier (non-preferred)brand drugs10.2 and 3-tier brand drugsaccumulated (PPIs)11. Abso-lute level change was 1.6(P<0.001)12. Abso-lute level change was -4.8 (P<0.001)13. Emergency room vis-its14. Absolutelevel change was 0.0011(P>0.05)15. Absolute levelchange was 0.001 (P>0.05)
W H A T ’ S N E W
Last assessed as up-to-date: 2 October 2007.
Date Event Description
18 March 2009 Amended Correction to typographical error.
88Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
21 August 2008 Amended Converted to new review format.
3 October 2007 New citation required and conclusions have changed Substantive amendment
C O N T R I B U T I O N S O F A U T H O R S
MOA, AA and ADO prepared the protocol. JPK and HS commented on protocol drafts. Two authors (of AA, MOA, GEV, JPK andHS) independenty reviewed all of the search results, abstracts and reference lists of relevant reports. AA, MOA and GEV applied theinclusion criteria, assessed the quality and extracted the data for the included studies. CR further developed quality criteria (based onthe Effective Practice and Organisation of Care (EPOC) criteria) for interrupted time series (ITS) and repeated measures studies andconducted the statistical reanalyses for the ITS studies, together with Åsa Vernby (ÅV) (those that were reanalysed). AA and MOAprepared the report. The others commented.
D E C L A R A T I O N S O F I N T E R E S T
MA has previously carried out short-term pharmaeconomic projects for the National Insurance Service and the Norwegian MedicinesAgency. From 1997-99 he worked for a private company, Brevreklame, doing market research for pharmaceutical firms in Norway.
HS is supported by the Dutch Health Care Insurance Board (CVZ).
JPK has previously worked one year each for the Danish Medicines Agency and Lundbeck A/S as part of a residency in clinicalpharmacology, and is currently employed five hours a week at the Danish Medicines Agency (Licensing Division) until January 2007.Since March 2007 he has been working 0-5 hours a week as advisor for Nordic Biotech.
S O U R C E S O F S U P P O R T
Internal sources
• Norwegian Knowledge Centre for the Health Services, Norway.• Health Services Research Unit, University of Aberdeen, UK.
89Pharmaceutical policies: effects of cap and co-payment on rational drug use (Review)
